Hydraulic turbine set valve fault diagnosis method, system, equipment and product

Abstract

The invention provides a hydraulic turbine set valve fault diagnosis method, system, equipment and product, and is applied to the field of hydroelectric equipment fault diagnosis. On one side of the edge end, an initial voiceprint signal, an initial vibration signal and temperature data of a hydraulic turbine set valve are collected; taking the initial voiceprint signal as a reference signal, and performing noise reduction processing on the initial vibration signal based on the time domain correlation between the fluid noise in the initial voiceprint signal and the first noise in the initial vibration signal; by taking the vibration noise reduction signal as a reference signal, performing noise reduction processing on the initial voiceprint signal based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial voiceprint signal; and based on the vibration noise reduction signal, the voiceprint noise reduction signal and the temperature data, performing fault diagnosis on the hydraulic turbine set valve to obtain a fault diagnosis result. Therefore, fault diagnosis is carried out through cooperative noise reduction of the voiceprint signal and the vibration signal in combination with multi-dimensional data, and the fault diagnosis accuracy of the hydraulic turbine set valve is improved.

Description

Technical Field

[0001] This application is applied to the field of fault diagnosis of hydropower equipment, and in particular relates to a method, system, equipment and product for fault diagnosis of turbine valves. Background Technology

[0002] In a hydroelectric power station, turbine units convert water energy into mechanical energy, which then drives generators to produce electricity. Therefore, the turbine unit is the core equipment of a hydroelectric power station, and its operational stability directly affects the power generation efficiency and safety of the power system.

[0003] As a critical component of hydroelectric turbine units, valves often suffer from energy loss, equipment damage, and even safety accidents due to faults such as internal leakage, jamming, and wear. Related technologies involve collecting vibration signals from turbine unit valves using vibration sensors, processing these signals through wavelet denoising or Kalman filtering, and then using the processed vibration signals to diagnose valve faults and obtain diagnostic results.

[0004] However, the above methods have low accuracy in diagnosing valve faults. Summary of the Invention

[0005] To address the aforementioned issues, this application proposes a method, system, equipment, and product for diagnosing valve faults in hydro turbine units, which can improve the efficiency of fault diagnosis for hydro turbine units.

[0006] The first aspect of this application provides a method for diagnosing valve faults in a hydro-turbine unit, applied to an edge end deployed on one side of the hydro-turbine unit, comprising: acquiring data from the hydro-turbine unit valve using sensor equipment to obtain an initial acoustic signature signal, an initial vibration signal, and temperature data; using the initial acoustic signature signal as a reference signal, performing noise reduction processing on the initial vibration signal based on the time-domain correlation between fluid noise in the initial acoustic signature signal and a first noise in the initial vibration signal to generate a vibration noise reduction signal, wherein the first noise is vibration noise caused by the transmission of fluid noise in the valve body structure; using the vibration noise reduction signal as a reference signal, performing noise reduction processing on the initial acoustic signature signal based on the frequency-domain correlation between mechanical resonance noise in the vibration noise reduction signal and a second noise in the initial acoustic signature signal to generate an acoustic signature noise reduction signal, wherein the second noise is noise transmitted from the mechanical resonance noise to the acoustic sensor through the valve body structure; and performing fault diagnosis on the hydro-turbine unit valve based on the vibration noise reduction signal, the acoustic signature noise reduction signal, and the temperature data to obtain a fault diagnosis result.

[0007] In one possible implementation, the step of using the initial acoustic signature signal as a reference signal and performing noise reduction processing on the initial vibration signal based on the characteristic correlation between fluid noise in the initial acoustic signature signal and a first noise in the initial vibration signal to generate a vibration noise-reduced signal includes: obtaining a first time-domain sequence from the initial acoustic signature signal according to a candidate frequency range corresponding to the fluid noise; obtaining a second time-domain sequence from the initial vibration signal according to the candidate frequency range corresponding to the first noise; performing correlation analysis on the first time-domain sequence and the second time-domain sequence based on the candidate lag time of the initial vibration signal relative to the initial acoustic signature signal to obtain an analysis result; and, if the analysis result indicates that the first time-domain sequence and the second time-domain sequence are correlated, performing noise reduction processing on the initial vibration signal according to the candidate lag time and the first time-domain sequence to generate the vibration noise-reduced signal.

[0008] In one possible implementation, the number of candidate lag times, the first time-domain sequence, and / or the second time-domain sequence is multiple, and the analysis result includes multiple correlation values. When the analysis result indicates that the first time-domain sequence is correlated with the second time-domain sequence, the initial vibration signal is denoised based on the candidate lag times and the first time-domain sequence to generate the denoised vibration signal. This includes: selecting the maximum correlation value among the multiple correlation values; determining that the time-domain sequence corresponding to the maximum correlation value in the first time-domain sequence is correlated with the time-domain sequence corresponding to the maximum correlation value in the second time-domain sequence when the maximum correlation value is greater than a set threshold; shifting the time-domain sequence corresponding to the maximum correlation value in the first time-domain sequence in the time domain based on the lag time corresponding to the maximum correlation value in the candidate lag times to obtain a third time-domain sequence; and using the third time-domain sequence as a reference signal, denoising the initial vibration signal through an adaptive filter to generate the denoised vibration signal.

[0009] In one possible implementation, the step of using the vibration noise reduction signal as a reference signal and performing noise reduction processing on the initial voiceprint signal based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial voiceprint signal to generate a voiceprint noise reduction signal includes: performing Fourier transform and resonance feature analysis on the vibration noise reduction signal to determine the mechanical resonance frequency of the vibration noise reduction signal; determining filter parameters based on the mechanical resonance frequency; and filtering the initial voiceprint signal using a notch filter or a comb filter based on the filter parameters to generate the voiceprint noise reduction signal.

[0010] In one possible implementation, determining the filter parameters based on the mechanical resonance frequency includes: determining the filter frequency based on the mechanical resonance frequency; and determining the filter bandwidth corresponding to the filter frequency based on the harmonic energy change of the vibration noise reduction signal at the mechanical resonance frequency; wherein the filter parameters include the filter frequency and the filter bandwidth.

[0011] In one possible implementation, before performing fault diagnosis on the turbine unit valve based on the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data, and obtaining the fault diagnosis result, the following operations are further included: using the vibration noise reduction signal as a reference signal, removing environmental noise from the acoustic noise reduction signal through an adaptive filter; and / or, performing temperature compensation on the vibration amplitude of the vibration noise reduction signal according to the temperature data and a pre-built temperature-vibration amplitude relationship model.

[0012] In one possible implementation, a pre-trained fault diagnosis model is deployed at the edge. The fault diagnosis of the turbine unit valves is performed based on the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data to obtain fault diagnosis results. This includes: acquiring the operating parameters of the turbine unit; and performing fault diagnosis of the turbine unit valves using the fault diagnosis model based on the vibration noise reduction signal, the acoustic noise reduction signal, the temperature data, and the operating parameters to obtain the fault diagnosis results.

[0013] In one possible implementation, the step of performing fault diagnosis on the turbine unit valve based on the vibration noise reduction signal, the acoustic noise reduction signal, the temperature data, and the operating parameters using the fault diagnosis model to obtain the fault diagnosis result includes: inputting the vibration noise reduction signal, the acoustic noise reduction signal, the temperature data, and the operating parameters into the fault diagnosis model; in the fault diagnosis model, analyzing the weight parameters corresponding to the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data based on the operating parameters; extracting features from the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data, and weighting the extracted features according to the weight parameters corresponding to the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data to obtain a fused feature vector; and performing fault diagnosis on the turbine unit valve based on the fused feature vector to obtain the fault diagnosis result.

[0014] In one possible implementation, the method further includes: sending first information to the cloud, the first information indicating the vibration noise reduction signal, the acoustic noise reduction signal, the temperature data, and the fault diagnosis result; receiving second information sent by the cloud, the second information indicating the updating of the acquisition parameters of the sensor device, the fault diagnosis frequency of the edge end, and / or the data upload frequency of the edge end.

[0015] A second aspect of this application provides a turbine generator unit valve fault diagnosis system, comprising: an edge end deployed on one side of the turbine generator unit; wherein the edge end is configured to perform the turbine generator unit valve fault diagnosis method as described in the first aspect of this application or any possible implementation thereof.

[0016] In one possible implementation, the turbine unit valve fault diagnosis system further includes a cloud platform; the cloud platform is configured to perform the following operations: receiving first information sent by the edge terminal, the first information indicating the vibration noise reduction signal of the turbine unit valve, the acoustic noise reduction signal of the turbine unit valve, the temperature data of the turbine unit valve, and the fault diagnosis result of the turbine unit valve; predicting the fault trend of the turbine unit valve using a fault trend prediction model based on the vibration noise reduction signal, the acoustic noise reduction signal, the temperature data, and the fault diagnosis result, and obtaining a fault trend prediction result; generating second information based on the fault trend prediction result, the second information indicating the updating of the sensor device's acquisition parameters, the fault diagnosis frequency of the edge terminal, and / or the data upload frequency of the edge terminal; and sending the second information to the edge terminal.

[0017] A third aspect of this application provides a fault diagnosis device for a turbine generator valve, applied at the edge end deployed on one side of the turbine generator, comprising: a data acquisition unit for acquiring data from the turbine generator valve using sensor equipment to obtain an initial acoustic signature signal, an initial vibration signal, and temperature data; a first noise reduction unit for using the initial acoustic signature signal as a reference signal and performing noise reduction processing on the initial vibration signal based on the time-domain correlation between fluid noise in the initial acoustic signature signal and a first noise in the initial vibration signal to generate a vibration noise reduction signal, wherein the first noise is vibration noise caused by the transmission of fluid noise in the valve body structure; a second noise reduction unit for using the vibration noise reduction signal as a reference signal and performing noise reduction processing on the initial acoustic signature signal based on the frequency-domain correlation between mechanical resonance noise in the vibration noise reduction signal and a second noise in the initial acoustic signature signal to generate an acoustic signature noise reduction signal, wherein the second noise is noise transmitted from the mechanical resonance noise to the acoustic sensor through the valve body structure; and a fault diagnosis unit for performing fault diagnosis on the turbine generator valve based on the vibration noise reduction signal, the acoustic signature noise reduction signal, and the temperature data to obtain a fault diagnosis result.

[0018] The fourth aspect of this application provides an electronic device, including a memory and a processor; the memory is connected to the processor and is used to store a program; the processor is used to implement the turbine unit valve fault diagnosis method as described in the first aspect of this application or any possible implementation of the first aspect of this application by running the program in the memory.

[0019] The fifth aspect of this application provides a chip including a processor and a data interface, wherein the processor reads and runs a program stored in a memory through the data interface to perform a turbine unit valve fault diagnosis method as described in the first aspect of this application or any possible implementation thereof.

[0020] The sixth aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the turbine unit valve fault diagnosis method as described in the first aspect of this application or any possible implementation thereof.

[0021] The seventh aspect of this application provides a storage medium storing a computer program, which, when executed by a processor, implements the turbine unit valve fault diagnosis method as described in the first aspect of this application or any possible implementation thereof.

[0022] According to the method, system, equipment, and product for diagnosing valve faults in a hydro-turbine unit proposed in this application, data is collected from the valves of the hydro-turbine unit via sensor equipment deployed at the edge end on one side of the unit, obtaining initial acoustic signature signals, initial vibration signals, and temperature signals. Utilizing the time-domain correlation between fluid noise in the initial acoustic signature signal and vibration noise caused by fluid noise transmitted through the valve body structure in the initial vibration signal, the initial vibration signal is denoised using the initial acoustic signature signal to obtain a denoised vibration signal. Utilizing the frequency-domain correlation between mechanical resonance noise in the denoised vibration signal and acoustic noise caused by mechanical resonance noise transmitted through the valve body structure in the initial acoustic signature signal, the initial acoustic signature signal is denoised using the denoised vibration signal to obtain a denoised acoustic signature signal. Thus, by utilizing the time-frequency domain correlation between noise in the acoustic signature signal and noise in the vibration signal, synergistic denoising of the acoustic signature and vibration signals is achieved, improving the denoising effect of both signals and accurately removing noise from them. Subsequently, based on vibration noise reduction signals, acoustic noise reduction signals, and temperature data, fault diagnosis was performed on the turbine unit valves, and the fault diagnosis results were obtained, which improved the accuracy of fault diagnosis of turbine unit valves. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in 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 embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the implementation environment involved in the embodiments of this application;

[0025] Figure 2 This is a flowchart illustrating the turbine unit valve fault diagnosis method provided in the embodiments of this application. Figure 1 ;

[0026] Figure 3 This is a flowchart illustrating the turbine unit valve fault diagnosis method provided in the embodiments of this application. Figure 2 ;

[0027] Figure 4 This is a flowchart illustrating the turbine unit valve fault diagnosis method provided in the embodiments of this application. Figure 3 ;

[0028] Figure 5 This is a schematic diagram of the structure of the turbine unit valve fault diagnosis device provided in the embodiments of this application. Figure 1 ;

[0029] Figure 6 This is a schematic diagram of the structure of the turbine unit valve fault diagnosis device provided in the embodiments of this application. Figure 2 ;

[0030] Figure 7 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] Fault diagnosis schemes for turbine generator valves may include: First, collecting single-sensor data from the turbine generator valves, such as vibration signals, acoustic signatures, or temperature data; then, denoising the single-sensor data, for example, using wavelet denoising or Kalman filtering to denoise the vibration signal; and finally, based on the denoised sensor data, performing fault diagnosis on the turbine generator valves, such as relying on high-frequency whistling characteristics for fault identification in acoustic signature recognition technology.

[0033] However, the operating environment of the turbine unit is a high-humidity and high-noise environment. The above-mentioned fault diagnosis schemes are not accurate enough in high-humidity and high-noise environments. On the one hand, single sensor data is easily interfered with by high noise, and high humidity may cause sensor drift. For example, water vapor condenses into water droplets in high humidity environments, and the water droplets cover the sensing element of the temperature sensor, causing the temperature sensor data to be distorted. Also, voiceprint recognition technology that relies on high-frequency whistling characteristics is easily affected by the environmental noise of water turbulence. On the other hand, in high-noise environments, the above-mentioned noise reduction methods are difficult to effectively preserve fault characteristics, and the noise reduction effect is limited, which cannot provide effective data support for accurate fault diagnosis.

[0034] To address the aforementioned issues, this application provides a method, system, equipment, and product for diagnosing faults in turbine generator unit valves. It collects acoustic signature signals, vibration signals, and temperature data from turbine generator unit valves. Utilizing the correlation between noise in the acoustic signature signal and noise in the vibration signal in the time-frequency domain, it performs collaborative noise reduction on both signals, improving the noise reduction effect. Then, by combining the noise-reduced acoustic signature signal, noise-reduced vibration signal, and temperature data, it achieves fault diagnosis of turbine generator unit valves based on multi-dimensional sensor data. By improving the noise reduction effect of the acoustic signature and vibration signals and utilizing multi-dimensional sensor data for fault diagnosis, this application effectively improves the accuracy of fault diagnosis for turbine generator unit valves.

[0035] Exemplary Implementation Environment

[0036] Please refer to Figure 1 , Figure 1 This is a schematic diagram of the implementation environment according to an embodiment of this application. The implementation environment according to this application includes an edge end 120 deployed on one side of the turbine unit 110.

[0037] Among them, the edge device 120 can be an edge computing device that connects to the sensor (it can be a server, a terminal, or a gateway device, such as an industrial-grade edge computing gateway, an embedded edge computing module, or an industrial IoT edge server), or a sensor device that integrates sensing and computing capabilities (such as a smart sensor).

[0038] Optional. The real-time environment involved in the embodiments of this application also includes a cloud 130, and the edge terminal 120 communicates with the cloud 130.

[0039] Exemplary methods

[0040] Please see Figure 2 In one exemplary embodiment, a method for diagnosing valve faults in a hydro-turbine unit is provided. On one side of the edge end, the method includes the following steps:

[0041] S201 uses sensor equipment to collect data from the turbine unit valves, obtaining initial acoustic signals, initial vibration signals, and temperature data.

[0042] The sensor devices include sound sensors, vibration sensors, and temperature sensors. For example, the sound sensor uses a microphone array, the vibration sensor uses an accelerometer, and the temperature sensor uses a contact temperature sensor.

[0043] Optionally, the edge device is an integrated acoustic, vibration, and temperature sensor that incorporates a sound sensor, a vibration sensor, and a temperature sensor, and has a built-in edge computing module. Thus, the edge device itself has the capability to acquire acoustic signals, vibration signals, and temperature data, and can also perform fault diagnosis of turbine unit valves locally based on this sensor data, improving the efficiency of turbine unit valve fault diagnosis.

[0044] In this embodiment, during the operation of the turbine unit valves: data can be periodically collected from the turbine unit valves using sensor equipment to obtain initial acoustic signature signals, initial vibration signals, and temperature data; alternatively, upon receiving a fault diagnosis command, data can be collected from the turbine unit valves using sensor equipment to obtain initial acoustic signature signals, initial vibration signals, and temperature data. The initial acoustic signature signals, initial vibration signals, and temperature data can serve as real-time sensing data for the turbine unit valves, enabling real-time diagnosis of the turbine unit valves.

[0045] Optionally, during the data acquisition process of the turbine unit valves, time stamp alignment technology is used to synchronize the sound sensor, vibration sensor and temperature sensor in time, so that the acquisition time error of the sound signature signal, vibration signal and temperature data is less than the error threshold.

[0046] For example, by using timestamp alignment technology, the sound sensor, vibration sensor, and temperature sensor are synchronized in time, so that the acquisition time error of the acoustic fingerprint signal, vibration signal, and temperature data is less than 100 microseconds (μs); by using a microphone array, the acoustic fingerprint signal of the turbine unit valve during operation is acquired at a sampling frequency of 48 kHz and a sampling depth of 16 bits to obtain the initial acoustic fingerprint signal; by using an accelerometer, the vibration signal of the turbine unit valve during operation is acquired at a sampling frequency of 10 kHz and a measurement range of ±5g to obtain the initial vibration signal; by using a contact temperature measurement module, the surface temperature of the turbine unit valve during operation is acquired at a sampling interval of 1 second and a resolution of 0.1℃ to obtain the temperature data.

[0047] S202, using the initial acoustic signature signal as a reference signal, and based on the time-domain correlation between the fluid noise in the initial acoustic signature signal and the first noise in the initial vibration signal, the initial vibration signal is denoised to generate a vibration denoising signal. The first noise is the vibration noise caused by the transmission of fluid noise in the valve body structure.

[0048] Fluid noise includes noise generated by water turbulence and cavitation. During the operation of the turbine unit, fluid noise generated by water turbulence and cavitation inside the valves propagates along the valve body structure as mechanical waves. Influenced by the mechanical filtering effect of the turbine unit valve body structure, it produces a corresponding resonance response, which is then captured by vibration sensors. For simplicity, the vibration noise caused by the transmission of fluid noise within the valve body structure is referred to as the primary noise. The primary noise and the fluid noise that causes it are from the same source and exhibit time-domain correlation. Since the primary noise is not caused by a turbine unit valve malfunction, it is necessary to remove it.

[0049] In this embodiment, based on the frequency range of the fluid noise distribution and the frequency range of the first noise distribution, a time-domain correlation analysis is performed on the acoustic fingerprint signal located in the frequency range of the fluid noise distribution in the initial acoustic fingerprint signal and the vibration signal located in the frequency range of the first noise distribution in the initial vibration signal. If it is determined that the acoustic fingerprint signal located in the frequency range of the fluid noise distribution and the vibration signal located in the frequency range of the first noise distribution are correlated in the time domain, the acoustic fingerprint signal located in the frequency range of the fluid noise distribution in the initial vibration signal is used as a reference signal to perform noise reduction processing on the initial vibration signal, generating a noise-reduced vibration signal. The frequency ranges of the fluid noise distribution and the first noise distribution can be determined by expert experience.

[0050] Therefore, by utilizing the temporal correlation of noise from the same source, vibration noise introduced by the valve body structure due to fluid noise can be accurately removed from the initial vibration signal. This avoids the accidental deletion of key fault features (such as vibration noise caused by valve wear or loosening) by traditional noise reduction methods, improves the accuracy of noise reduction of the initial vibration signal, provides accurate data support for subsequent turbine unit valve fault identification, and improves the accuracy of fault identification.

[0051] S203, using the vibration noise reduction signal as a reference signal, and based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial acoustic pattern signal, performs noise reduction processing on the initial acoustic pattern signal to generate an acoustic pattern noise reduction signal. The second noise is the noise transmitted to the acoustic sensor through the valve body structure by the mechanical resonance noise.

[0052] Mechanical resonance noise refers to the noise generated during the operation of a turbine unit due to the resonance of mechanical components, such as harmonic noise caused by valve jamming. Mechanical resonance noise propagates along the valve body structure as mechanical waves. When the vibration reaches the valve body surface, it causes periodic changes in the air density near the surface, forming sound waves that are captured by acoustic sensors and manifest as low-frequency noise in the acoustic signature. For simplicity, the noise transmitted from the mechanical resonance noise through the valve body structure to the acoustic sensor is called the second noise. The second noise is noise in the same frequency band as the mechanical resonance noise, and the two are correlated in the frequency domain, i.e., they are coupled in the frequency domain. Since the second noise is not caused by a turbine unit valve malfunction, it is necessary to remove it.

[0053] In this embodiment, the mechanical resonance frequency of the vibration signal can be analyzed. Since the mechanical resonance noise in the vibration noise reduction signal is frequency-domain correlated with the second noise in the initial acoustic pattern signal, the acoustic pattern signal of the corresponding frequency in the initial acoustic pattern signal can be filtered based on the mechanical resonance frequency to remove the second noise in the initial acoustic pattern signal and obtain the acoustic pattern noise reduction signal.

[0054] Therefore, by utilizing the frequency domain correlation between mechanical resonance noise and the second noise, the noise introduced by mechanical resonance in the initial acoustic signature signal can be targeted and extracted. This avoids the accidental deletion of key fault features by traditional noise reduction methods, improves the accuracy of noise reduction of the initial acoustic signature signal, provides accurate data support for subsequent turbine unit valve fault identification, and improves the accuracy of fault identification.

[0055] S204, based on vibration noise reduction signal, acoustic noise reduction signal and temperature data, performs fault diagnosis on the turbine unit valves and obtains fault diagnosis results.

[0056] Among them, the fault diagnosis results indicate whether the turbine unit valves have malfunctioned.

[0057] Optionally, the fault diagnosis results indicate the fault type and / or fault severity of the turbine unit valves.

[0058] Furthermore, the failure types of turbine unit valves may include one or more of the following: internal leakage, valve jamming, and valve wear. These failure types can be predicted by abnormal changes in vibration noise reduction signals, acoustic noise reduction signals, and / or temperature data.

[0059] In this embodiment, after obtaining the vibration noise reduction signal and the acoustic noise reduction signal, further optimization processing can be performed on the vibration noise reduction signal and the acoustic noise reduction signal. Alternatively, the vibration noise reduction signal, the acoustic noise reduction signal and the temperature data can be directly combined to perform turbine unit valve fault diagnosis based on multi-dimensional sensor data, and finally obtain the fault diagnosis effect of turbine unit valve.

[0060] In this embodiment, the time-domain correlation between fluid noise in the acoustic signature signal and the first noise in the vibration signal, and the frequency-domain correlation between mechanical noise in the vibration signal and the second noise in the acoustic signature signal, are used to perform collaborative noise reduction on the acoustic signature signal and the vibration signal. This avoids the accidental deletion of key fault features by traditional noise reduction methods and improves the accuracy of noise reduction for both acoustic signature and vibration signals. Furthermore, by combining the noise-reduced acoustic signature signal, the noise-reduced vibration signal, and temperature data, fault diagnosis of turbine generator valves based on multi-dimensional sensor data is achieved. Therefore, the accuracy of fault diagnosis for turbine generator valves is effectively improved.

[0061] Below, some possible implementation methods are provided for some steps in the above embodiments.

[0062] In one possible implementation, using the initial acoustic signature signal as a reference signal, the initial vibration signal is denoised based on the temporal correlation between the fluid noise in the initial acoustic signature signal and the first noise in the initial vibration signal to generate a denoised vibration signal, i.e., S202. This may include: obtaining a first time-domain sequence from the initial acoustic signature signal according to the candidate frequency range corresponding to the fluid noise; obtaining a second time-domain sequence from the initial vibration signal according to the candidate frequency range corresponding to the first noise; performing correlation analysis on the first time-domain sequence and the second time-domain sequence based on the candidate lag time of the initial vibration signal relative to the initial acoustic signature signal to obtain the analysis result; and, if the analysis result indicates that the first time-domain sequence and the second time-domain sequence are correlated, denoising the initial vibration signal according to the candidate lag time and the first time-domain sequence to generate a denoised vibration signal. Thus, through the above analysis and denoising process, the first noise in the initial vibration signal is removed in a targeted manner, avoiding false suppression of key fault features in the initial vibration signal and improving the denoising effect of the initial vibration signal.

[0063] The first time-domain sequence is the signal sequence in the initial acoustic signature signal that falls within the candidate frequency range corresponding to the fluid noise. The second time-domain sequence is the signal sequence in the initial vibration signal that falls within the candidate frequency range corresponding to the first noise. The candidate frequency ranges corresponding to the fluid noise and the first noise can be determined based on expert experience.

[0064] As an example, the candidate frequency range for fluid noise is 2kHz to 5kHz, and the candidate frequency range for the first noise is 0.5kHz to 1kHz.

[0065] The candidate lag time of the initial vibration signal relative to the initial acoustic signature signal can be determined based on the time length of the initial vibration signal and the time length of the initial acoustic signature signal.

[0066] In this embodiment, the initial acoustic signature signal can be filtered using a bandpass filter corresponding to the candidate frequency range of the fluid noise to obtain a first time-domain sequence. Similarly, the initial vibration signal can be filtered using a bandpass filter corresponding to the candidate frequency range of the first noise to obtain a second time-domain sequence. Preprocessing of the first and second time-domain sequences can improve the accuracy of subsequent time-domain correlation analysis. Next, the first time-domain sequence can be shifted in the time domain according to the candidate lag time of the initial vibration signal relative to the initial acoustic signature signal to obtain a shifted first time-domain sequence. Correlation calculation is then performed on the shifted first and second time-domain sequences. Alternatively, the second time-domain sequence can be shifted in the time domain according to the candidate lag time of the initial vibration signal relative to the initial acoustic signature signal to obtain a shifted second time-domain sequence. Correlation calculation is then performed on the first and shifted second time-domain sequences to finally obtain the correlation value between the first and second time-domain sequences. Therefore, the analysis results can include the correlation value between the first and second time-domain sequences. The correlation between the first and second time-domain sequences can be determined based on their correlation values. If they are correlated, the initial vibration signal can be denoised using the candidate lag time and the first time-domain sequence to generate a denoised vibration signal. During this process, the first time-domain sequence can be shifted in the time domain according to the candidate lag time to obtain a shifted first time-domain sequence. This shifted first time-domain sequence is then used as a reference signal to further denoise the initial vibration signal, generating the denoised vibration signal.

[0067] Optionally, preprocessing the first and second time-domain sequences may include: normalizing both sequences, with the amplitude range of the normalized first and second time-domain sequences being 0–1; considering that both sequences are non-stationary, the normalized sequences are first framed to obtain framed sequences, improving their continuity; then, a window function is used to apply windowing to the signal frames in both sequences, reducing edge abrupt changes caused by frame processing, lowering spectral leakage, and making the sequences smoother. Thus, normalization and framed windowing significantly improve the signal quality of both sequences, making it easier to accurately calculate their temporal correlation.

[0068] Furthermore, the window function can be a Hamming window, which can effectively reduce spectral leakage.

[0069] Optionally, there may be multiple candidate lag times, first time-domain sequences, and / or second time-domain sequences, and the analysis results may include multiple correlation values.

[0070] In one example, during the correlation calculation of sequence segments of the first and second time-domain sequences at corresponding time periods based on candidate lag times of the initial vibration signal relative to the initial acoustic signature signal, multiple candidate lag times can be traversed. Based on the currently traversed candidate lag time, either the first or second time-domain sequence is shifted in the time domain, and then the correlation between the first and second time-domain sequences is calculated again to obtain their correlation values. By traversing multiple candidate lag times, multiple correlation values ​​can be obtained.

[0071] In another example, there are multiple candidate frequency ranges corresponding to the fluid noise. Multiple first time-domain sequences are obtained from the initial acoustic signature signal according to these multiple candidate frequency ranges. And / or, there are multiple candidate frequency ranges corresponding to the first noise. Multiple second time-domain sequences are obtained from the initial vibration signal according to these multiple candidate frequency ranges. When there are multiple first time-domain sequences and / or multiple second time-domain sequences, multiple correlation calculations can be performed based on the first and second time-domain sequences to obtain multiple correlation values.

[0072] Optionally, the candidate frequency range corresponding to the fluid noise has a corresponding relationship with the candidate frequency range corresponding to the first noise. Based on this correspondence, multiple time-domain sequence pairs can be constructed, each of which includes a first time-domain sequence and a second time-domain sequence. Multiple candidate lag times are traversed, and the correlation is calculated for each time-domain sequence pair according to the traversed candidate lag times, finally obtaining multiple correlation values.

[0073] For example, taking a frequency range width of 1 kHz as an example, the multiple candidate frequency ranges corresponding to fluid noise include: 0 kHz to 1 kHz, 1 kHz to 2 kHz, 2 kHz to 3 kHz, ...; taking a frequency range width of 0.5 kHz as an example, the multiple candidate frequency ranges corresponding to the first noise include: 0 kHz to 0.5 kHz, 0.5 kHz to 1 kHz, ... By traversing these frequency ranges and extracting time-domain sequences according to the traversed frequency ranges, multiple first time-domain sequences and multiple second time-domain sequences can be obtained.

[0074] Optionally, the formula for calculating the temporal correlation between the first time-domain sequence and the second time-domain sequence is as follows:

[0075]

[0076] Where τ represents the candidate lag time, x(n) represents the first time-domain sequence, y(n+τ) represents the second time-domain sequence shifted according to the candidate lag time, and R xy (τ) represents the correlation value between the first time-domain sequence and the second time-domain sequence in the time domain.

[0077] Optionally, among multiple correlation values, the largest correlation value is selected; if the largest correlation value is greater than a set threshold, the correlation between the time-domain sequence corresponding to the largest correlation value in the first time-domain sequence and the time-domain sequence corresponding to the largest correlation value in the second time-domain sequence is determined; based on the lag time corresponding to the largest correlation value among the candidate lag times, the time-domain sequence corresponding to the largest correlation value in the first time-domain sequence is shifted in the time domain to obtain a third time-domain sequence; using the third time-domain sequence as a reference signal, the initial vibration signal is denoised through an adaptive filter to generate a vibration denoised signal.

[0078] Wherein, the time-domain sequence corresponding to the maximum correlation value in the first time-domain sequence refers to the first time-domain sequence that participated in the calculation of the maximum correlation value; the time-domain sequence corresponding to the maximum correlation value in the second time-domain sequence refers to the second time-domain sequence that participated in the calculation of the maximum correlation value; and the lag time corresponding to the maximum correlation value in the candidate lag time refers to the lag time that participated in the calculation of the maximum correlation value.

[0079] For example, based on multiple candidate lag times, correlation values ​​are calculated between multiple first time-domain sequences and multiple second time-domain sequences for each pair of time-domain sequences, resulting in multiple correlation values. The first time-domain sequence that yields the largest correlation value is selected from the multiple first time-domain sequences; the second time-domain sequence that yields the largest correlation value is selected from the multiple second time-domain sequences; and the first time-domain sequence that yields the largest correlation value is selected from the multiple candidate lag times.

[0080] Here, a threshold is set, for example, 0.6, if the maximum R xy If (τ) is greater than 0.6, then R is considered to have the largest participation. xy (τ) is related to the first and second time-domain sequences calculated.

[0081] In this optional approach, the third time-domain sequence and the initial vibration signal can be input together into an adaptive filter. The adaptive filter performs noise reduction processing on the initial vibration signal with reference to the third time-domain sequence to generate a noise-reduced vibration signal. Thus, by utilizing the temporal correlation between fluid noise and the first noise, correlation analysis is performed on the time-domain sequence of the initial acoustic signature signal in the frequency range of fluid noise and the time-domain sequence of the initial vibration signal in the frequency range of the first noise. This identifies the time-domain sequence in the initial acoustic signature signal that is related to the initial vibration signal. Referring to this time-domain sequence and the lag time of the initial vibration signal relative to the initial acoustic signature signal, targeted noise reduction is performed on the initial vibration signal, improving the accuracy of removing the first noise from the initial vibration signal.

[0082] Optionally, if the maximum correlation value is less than a set threshold, the initial vibration signal will not be denoised based on the time-domain correlation between the fluid noise in the initial acoustic signal and the first noise in the initial vibration signal, thus avoiding false suppression of the fault-related signal in the initial vibration signal.

[0083] In one possible implementation, using the vibration noise reduction signal as a reference signal, the initial acoustic signature signal is denoised based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial acoustic signature signal to generate an acoustic signature denoised signal, i.e., S203. This includes: performing Fourier transform and resonance feature analysis on the vibration noise reduction signal to determine the mechanical resonance frequency of the vibration noise reduction signal; determining filter parameters based on the mechanical resonance frequency; and filtering the initial acoustic signature signal using a notch filter or a comb filter based on the filter parameters to generate the acoustic signature denoised signal. On the one hand, Fourier transform and resonance feature analysis can improve the accuracy of determining the mechanical resonance frequency of the vibration noise reduction signal. On the other hand, based on the mechanical resonance frequency, more targeted filter parameters can be determined for filtering the initial acoustic signal. By combining these filter parameters, notch filters or comb filters can accurately remove the second noise caused by mechanical resonance noise from the initial acoustic signal, retain the acoustic characteristics of unrelated frequency bands (such as the 200Hz to 300Hz bubble sound generated by valve internal leakage faults), and avoid false suppression of fault-related signals in the initial acoustic signal.

[0084] Among them, the Fourier transform can be the Fast Fourier Transform (FFT).

[0085] Optionally, filter parameters are determined based on the mechanical resonant frequency, including: determining the filter frequency based on the mechanical resonant frequency; and determining the filter bandwidth corresponding to the filter frequency based on the harmonic energy variation of the vibration noise reduction signal at the mechanical resonant frequency. The filter parameters include both the filter frequency and the filter bandwidth. On one hand, considering the frequency domain correlation between mechanical operating noise and the second noise, a filter frequency related to the mechanical resonant frequency is used, or the mechanical resonant frequency is used as the filter frequency, enabling the notch filter or comb filter to filter the second noise from the initial acoustic signature signal. On the other hand, adapting to the harmonic energy variation of the vibration noise reduction signal at the mechanical resonant frequency allows for dynamic adjustment of the filter bandwidth, avoiding excessive suppression of the true fault characteristics in the initial acoustic signature signal.

[0086] In this optional approach, a mapping relationship between harmonic energy variation and filter bandwidth can be established in advance. After determining the filter frequency, based on the harmonic energy variation of the vibration noise reduction signal at the mechanical resonant frequency, the corresponding filter bandwidth is found in this mapping relationship, and this filter bandwidth is determined as the filter bandwidth corresponding to the filter frequency, i.e., the filter bandwidth of the notch filter or comb filter.

[0087] Please see Figure 3 In one exemplary embodiment, a method for diagnosing valve faults in a hydro-turbine unit is provided. On one side of the edge end, the method includes the following steps:

[0088] S301 collects data from the turbine unit valves using sensor equipment, obtaining initial acoustic signals, initial vibration signals, and temperature data.

[0089] S302, using the initial acoustic signature signal as a reference signal, performs noise reduction processing on the initial vibration signal based on the time-domain correlation between the fluid noise in the initial acoustic signature signal and the first noise in the initial vibration signal, and generates a vibration noise reduction signal.

[0090] S303 uses the vibration noise reduction signal as a reference signal and performs noise reduction processing on the initial voiceprint signal based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial voiceprint signal to generate a voiceprint noise reduction signal.

[0091] The implementation principles and technical effects of S301 to S303 are the same as those in the aforementioned embodiments and will not be repeated here.

[0092] S304 uses the vibration noise reduction signal as a reference signal and removes environmental noise from the acoustic noise reduction signal through an adaptive filter.

[0093] S305 performs temperature compensation on the vibration amplitude of the vibration noise reduction signal based on temperature data and a pre-built temperature-vibration amplitude relationship model.

[0094] S304 and S305 are optional steps. You can choose not to execute S304 and S305, execute one of S304 and S305, or execute both S304 and S305.

[0095] In S304, after the initial voiceprint signal is denoised as described above to obtain the voiceprint denoised signal, the voiceprint denoised signal may still contain a large amount of environmental noise. The vibration denoised signal and the voiceprint denoised signal can be input into an adaptive filter. The adaptive filter refers to the vibration denoised signal and performs adaptive filtering on the voiceprint denoised signal to suppress the environmental noise in the voiceprint denoised signal.

[0096] In S305, under high humidity conditions, the change in the thermal expansion coefficient of the valve sealing material may affect the amplitude of the vibration signal. Temperature compensation can be performed on the vibration amplitude of the vibration noise reduction signal by combining temperature data and a pre-built temperature-vibration amplitude relationship model, thereby eliminating the interference of ambient temperature on the extraction of valve fault features from the vibration signal and improving the accuracy of fault diagnosis of turbine unit valves.

[0097] In the process of pre-constructing the temperature vibration amplitude relationship model, the vibration amplitude of the turbine unit valves under different temperature environments can be obtained. Based on the vibration amplitude of the turbine unit valves under different temperature environments, the relationship between ambient temperature and vibration amplitude is fitted to obtain the temperature vibration amplitude relationship model. For example, the least squares method can be used to fit the relationship between ambient temperature and vibration amplitude to obtain the temperature vibration amplitude relationship model.

[0098] In the process of temperature compensation for the vibration amplitude of the vibration noise reduction signal by combining temperature data and a pre-built temperature vibration amplitude relationship model, the vibration amplitude under the temperature data can be determined by the temperature vibration amplitude relationship model, the vibration amplitude under the standard temperature can be determined by the temperature vibration amplitude relationship model, the vibration amplitude difference corresponding to the vibration noise reduction signal can be obtained based on the vibration amplitude under the temperature data and the vibration amplitude under the standard temperature, and the vibration amplitude of the vibration noise reduction signal can be compensated for temperature according to the vibration amplitude difference.

[0099] Optionally, the formula for calculating the vibration amplitude difference is expressed as follows:

[0100] ΔA=f(T r H r )-f(T l H l )

[0101] Where ΔA represents the vibration amplitude difference, f() represents the temperature vibration amplitude relationship model, and T r This represents the actual temperature, i.e., the temperature data acquired by the sensor device in this embodiment, H. r T representsr The corresponding vibration amplitude, T l H represents the standard temperature (e.g., 25 degrees Celsius). l T represents l The corresponding vibration amplitude.

[0102] The temperature compensation formula is expressed as:

[0103] A′=A raw -ΔA

[0104] Among them, A raw A represents the vibration amplitude of the vibration noise reduction signal, and A′ represents the vibration amplitude of the vibration noise reduction signal after temperature compensation.

[0105] S306, based on vibration noise reduction signal, acoustic noise reduction signal and temperature data, performs fault diagnosis on the turbine unit valves and obtains fault diagnosis results.

[0106] The implementation principle and technical effects of S306 are the same as those in the aforementioned embodiments, and will not be repeated here.

[0107] Optionally, a pre-trained fault diagnosis model can be deployed at the edge. This fault diagnosis model is a deep learning model capable of processing multi-dimensional data. Therefore, based on multi-dimensional sensor data such as vibration noise reduction signals, acoustic noise reduction signals, and temperature data, the deep learning model can be further utilized as a fault diagnosis model to improve the accuracy of fault diagnosis for turbine unit valves.

[0108] Furthermore, the fault diagnosis model is a lightweight deep learning model, which improves the efficiency of fault diagnosis of turbine unit valves and reduces the fault diagnosis response time at the edge.

[0109] In one possible implementation, S306 includes: S3061, acquiring the operating parameters of the turbine unit; S3062, based on vibration noise reduction signals, acoustic noise reduction signals, temperature data, and operating parameters, performing fault diagnosis on the turbine unit valves using a fault diagnosis model to obtain fault diagnosis results. This implementation takes into account that the fault characteristics of the same fault may differ under different operating conditions using vibration noise reduction signals, acoustic noise reduction signals, and temperature data. For example, the acoustic characteristics of valve leakage under high head conditions are significantly different from those under low head conditions. Under high head conditions, the acoustic characteristics of valve leakage are high-frequency whistling (frequency > 5kHz), while under low head conditions, the acoustic characteristics are mid-frequency noise (frequency range 1kHz to 3kHz). By combining multi-dimensional sensing data with operating parameters, the accuracy of fault diagnosis of turbine unit valves is improved.

[0110] Operating parameters may include (the height of the water flow), flow rate, power, etc.

[0111] In S3061, the operating condition parameters sent by the operating condition function module in the turbine unit can be obtained, which can indicate the real-time operating condition of the turbine unit.

[0112] In one possible implementation, S3062 includes: inputting the vibration noise reduction signal, acoustic noise reduction signal, temperature data, and operating parameters into a fault diagnosis model; in the fault diagnosis model, analyzing the weight parameters corresponding to the vibration noise reduction signal, acoustic noise reduction signal, and temperature data based on the operating parameters; extracting features from the vibration noise reduction signal, acoustic noise reduction signal, and temperature data, and weighting the extracted features according to the weight parameters corresponding to the vibration noise reduction signal, acoustic noise reduction signal, and temperature data to obtain a fused feature vector; and performing fault diagnosis on the turbine unit valves based on the fused feature vector to obtain a fault diagnosis result. This implementation achieves two main benefits: First, it adapts to the operating parameters of the turbine unit, enabling real-time adjustment of the weight parameters corresponding to vibration noise reduction signals, acoustic noise reduction signals, and temperature data. This allows for dynamic weighted feature fusion of multi-dimensional sensor data, selectively amplifying the most identifiable feature signals under real-time operating conditions, effectively suppressing irrelevant interference, and making the fused feature vector of vibration noise reduction signals, acoustic noise reduction signals, and temperature data more closely match the actual operating characteristics of the turbine unit. Second, it strengthens the correlation between multi-dimensional sensor data and operating parameters, enabling the fault diagnosis model to more accurately capture the intrinsic relationship between changes in the turbine unit's operating conditions and the sensor data of the turbine unit's valves. This improves the fault diagnosis model's adaptability to different operating conditions of the turbine unit, thereby effectively improving the accuracy of fault diagnosis for turbine unit valves.

[0113] In this implementation, the fault diagnosis model performs feature extraction and feature concatenation on the vibration noise reduction signal, acoustic noise reduction signal, and temperature data to obtain an original feature vector. Based on operating parameters, the weight parameters corresponding to the vibration noise reduction signal, acoustic noise reduction signal, and temperature data are analyzed. According to these weight parameters, the feature elements in the original feature vector corresponding to the vibration noise reduction signal, acoustic noise reduction signal, and temperature data are weighted to obtain a fused feature vector. Finally, based on the fused feature vector, fault diagnosis is performed on the turbine unit valves to obtain the fault diagnosis result.

[0114] Optionally, a weight mapping relationship can be pre-constructed, including weight parameters corresponding to vibration signals, acoustic signatures, and temperature data under various reference operating conditions. During the analysis of the weight parameters corresponding to the vibration-denoised signal, acoustic signature, and temperature data based on the operating conditions, the weight parameters can be found in the weight mapping table for each operating condition. Thus, the weight mapping relationship provides an accurate reference for determining the weight parameters of the vibration-denoised signal, acoustic signature, and temperature data under the operating conditions, improving the accuracy of feature fusion through multi-dimensional sensor data.

[0115] In this embodiment, after the initial acoustic signature signal and initial vibration signal are denoised collaboratively to obtain the acoustic signature denoised signal and vibration denoised signal, the vibration denoised signal is used as a reference signal. An adaptive filter is used to remove environmental noise from the acoustic signature denoised signal, and / or, based on temperature data and a pre-built temperature-vibration amplitude relationship model, temperature compensation is performed on the vibration amplitude of the vibration denoised signal to further improve the accuracy of the acoustic signature denoised signal and / or vibration denoised signal. During fault diagnosis, a fault diagnosis model and / or operating condition parameters are introduced, effectively improving the accuracy of fault diagnosis for turbine unit valves.

[0116] Please see Figure 4 In one exemplary embodiment, a method for diagnosing valve faults in a hydro-turbine unit is provided, the method comprising the following steps:

[0117] S401, the edge end collects data from the turbine unit valves through sensor equipment to obtain initial acoustic signal, initial vibration signal and temperature data;

[0118] S402, the edge end uses the initial acoustic signature signal as a reference signal, and performs noise reduction processing on the initial vibration signal based on the time-domain correlation between the fluid noise in the initial acoustic signature signal and the first noise in the initial vibration signal to generate a vibration noise reduction signal.

[0119] S403, at the edge end, the vibration noise reduction signal is used as a reference signal. Based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial voiceprint signal, the initial voiceprint signal is denoised to generate a voiceprint noise reduction signal.

[0120] S404, based on vibration noise reduction signal, acoustic noise reduction signal and temperature data, performs fault diagnosis on the turbine unit valves and obtains the fault diagnosis results.

[0121] The implementation principles and technical effects of S401 to S404 are the same as those in the aforementioned embodiments and will not be repeated here.

[0122] S405, the edge device sends the first information to the cloud, the first information indicating vibration noise reduction signal, soundprint noise reduction signal, temperature data and fault diagnosis results.

[0123] The first piece of information may include vibration noise reduction signal, acoustic noise reduction signal, temperature data, and fault diagnosis results.

[0124] In this embodiment, after the edge device completes the fault diagnosis of the turbine unit valve, it sends the first information to the cloud to realize the reporting of vibration noise reduction signal, acoustic noise reduction signal, temperature data and fault diagnosis results. On the one hand, the cloud can record this information, especially the fault diagnosis results; on the other hand, the cloud can also perform some other processing based on this information.

[0125] Optionally, if the fault diagnosis result indicates that the turbine unit valve has malfunctioned, the edge terminal sends the first information to the cloud, thereby promptly sending relevant data on the turbine unit valve malfunction to the cloud, so that the cloud can take timely measures to address the turbine unit valve malfunction.

[0126] S406, based on vibration noise reduction signals, acoustic noise reduction signals, temperature data and fault diagnosis results, the cloud uses a fault trend prediction model to predict the fault trend of the turbine unit valves and obtain the fault trend prediction results.

[0127] Among them, a pre-trained fault trend prediction model is deployed in the cloud. The fault trend prediction model has the ability to predict the fault evolution trend of turbine valves. Its training data can include vibration signals, acoustic signals and temperature data of turbine valves under various fault evolution trends.

[0128] Optionally, the fault trend prediction model is a deep learning model combining a deep belief network (DBN) and a long short-term memory (LSTM) network, which can be called an LSTM-DBN hybrid model. This model possesses the ability to handle complex data with high dimensionality, nonlinearity, and time dependence. In the fault trend prediction model, LSTM can be used to perform time-domain sequence analysis on vibration noise reduction signals, acoustic signature noise reduction signals, and temperature data to extract corresponding time-series features, obtaining the LSTM output data. The LSTM output data is then input into the DBN, where a restricted Boltzmann machine (RBM) is used to extract abstract high-order features layer by layer. Based on these high-order features, the evolution trend of turbine unit valve faults is predicted. In this optional approach, LSTM is used to handle time dependence, and DBN is used to mine high-dimensional feature correlations across sensors, improving the accuracy of the fault trend prediction model in predicting the evolution trend of turbine unit valve faults.

[0129] S407, based on the fault trend prediction results, the cloud generates second information, which instructs the sensor device's acquisition parameters, the fault diagnosis frequency at the edge, and / or the data upload frequency at the edge to be updated.

[0130] Among them, the fault trend prediction results indicate the fault evolution trend of turbine unit valves. The fault diagnosis frequency at the edge refers to the frequency with which the edge performs fault diagnosis on turbine unit valves, which can also be reflected as the fault diagnosis cycle. The data upload frequency at the edge refers to the frequency at which the edge sends initial information to the cloud. The fault diagnosis frequency and the data upload frequency can be the same or different.

[0131] In one example, the cloud can determine, based on the fault evolution trend indicated by the fault trend prediction results, to accelerate the sensor device's acquisition parameters, the fault diagnosis frequency at the edge, and / or the data upload frequency at the edge. In this case, the second information indicates that the sensor device's acquisition parameters, the fault diagnosis frequency at the edge, and / or the data upload frequency at the edge should be accelerated. Alternatively, it can determine to slow down the sensor device's acquisition parameters, the fault diagnosis frequency at the edge, and / or the data upload frequency at the edge. In this case, the second information indicates that the sensor device's acquisition parameters, the fault diagnosis frequency at the edge, and / or the data upload frequency at the edge should be slowed down. Or, it can determine to keep the sensor device's acquisition parameters, the fault diagnosis frequency at the edge, and / or the data upload frequency at the edge unchanged. In this case, the second information may not be generated, or a second information indicating that the sensor device's acquisition parameters, the fault diagnosis frequency at the edge, and / or the data upload frequency at the edge should be kept unchanged may be generated.

[0132] For example, the cloud can predict the spread rate of internal leakage faults in turbine valves. When the predicted spread rate is greater than a threshold, the data upload frequency and fault diagnosis frequency at the edge can be accelerated.

[0133] In another example, the cloud can adjust the acquisition parameters of the sensor device, the fault diagnosis frequency of the edge end, and / or the data upload frequency of the edge end according to the fault evolution trend indicated by the fault trend prediction results, to obtain the updated values ​​corresponding to the acquisition parameters, the updated values ​​corresponding to the fault diagnosis frequency, and / or the updated values ​​corresponding to the data upload frequency. The second information includes the updated values ​​corresponding to the acquisition parameters, the updated values ​​corresponding to the fault diagnosis frequency, and / or the updated values ​​corresponding to the data upload frequency.

[0134] S408, the cloud sends a second message to the edge.

[0135] In this embodiment, the cloud sends second information to the edge device. The edge device can update the acquisition parameters of the sensor device, the fault diagnosis frequency of the edge device, and / or the data upload frequency of the edge device based on the second information, thereby improving the rationality of the acquisition parameters of the sensor device, the fault diagnosis frequency of the edge device, and / or the data upload frequency of the edge device.

[0136] In this embodiment, by collaborating between the edge and cloud, and combining the failure evolution trend of turbine valves, the failure diagnosis strategy for turbine valves on the edge side is flexibly adjusted to improve the overall efficiency of turbine valve failure diagnosis. For example, when the failure evolution trend of turbine valves is relatively optimistic, the resource consumption of the edge is reduced; when the failure evolution trend of turbine valves is not optimistic, the data acquisition, failure diagnosis, and / or data upload of the edge are accelerated to achieve refined real-time monitoring of turbine valves.

[0137] Exemplary systems and apparatus

[0138] Accordingly, this application also provides a turbine unit valve fault diagnosis system.

[0139] In an exemplary embodiment, a turbine generator valve fault diagnosis system provided in this application includes an edge end, which is deployed on one side of the turbine generator and configured to perform the turbine generator valve fault diagnosis method located on one side of the edge end provided in any of the above embodiments.

[0140] Optionally, the turbine unit valve fault diagnosis system also includes a cloud platform. The cloud platform is configured to perform the following operations: receive first information sent from the edge terminal, the first information indicating the vibration noise reduction signal, acoustic noise reduction signal, temperature data, and fault diagnosis results of the turbine unit valve; predict the fault trend of the turbine unit valve using a fault trend prediction model based on the vibration noise reduction signal, acoustic noise reduction signal, temperature data, and fault diagnosis results, obtaining a fault trend prediction result; generate second information based on the fault trend prediction result, the second information indicating the updating of the sensor equipment's acquisition parameters, the edge terminal's fault diagnosis frequency, and / or the edge terminal's data upload frequency; and send the second information to the edge terminal.

[0141] The turbine unit valve fault diagnosis system provided in this embodiment belongs to the same concept as the turbine unit valve fault diagnosis method provided in any of the above embodiments of this application. It can execute the turbine unit valve fault diagnosis method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the turbine unit valve fault diagnosis method. Technical details not described in detail in this embodiment can be found in the content of the corresponding method embodiments of this application, and will not be repeated here.

[0142] Accordingly, this application also provides a valve fault diagnosis device for a water turbine unit.

[0143] Please see Figure 5 In one exemplary embodiment, a turbine generator valve fault diagnosis device 500 is provided, applied to the edge end deployed on one side of the turbine generator. The turbine generator valve fault diagnosis device 500 includes: a data acquisition unit 501, a first noise reduction unit 502, a second noise reduction unit 503, and a fault diagnosis unit 504. Wherein:

[0144] The data acquisition unit 501 is used to acquire data from the turbine generator valves using sensor devices, obtaining initial acoustic signature signals, initial vibration signals, and temperature data. The first noise reduction unit 502 is used to perform noise reduction processing on the initial vibration signal based on the time-domain correlation between the fluid noise in the initial acoustic signature signal and the first noise in the initial vibration signal, using the initial acoustic signature signal as a reference signal, to generate a vibration noise reduction signal. The first noise is the vibration noise caused by the transmission of fluid noise in the valve body structure. The second noise reduction unit 503 is used to perform noise reduction processing on the initial acoustic signature signal based on the frequency-domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial acoustic signature signal, using the vibration noise reduction signal as a reference signal, to generate an acoustic signature noise reduction signal. The second noise is the noise transmitted from the mechanical resonance noise to the acoustic sensor through the valve body structure. The fault diagnosis unit 504 is used to perform fault diagnosis on the turbine generator valves based on the vibration noise reduction signal, the acoustic signature noise reduction signal, and the temperature data, obtaining the fault diagnosis result.

[0145] In one possible implementation, the first noise reduction unit 502 is specifically used to: obtain a first time-domain sequence from the initial acoustic signature signal according to the candidate frequency range corresponding to the fluid noise; obtain a second time-domain sequence from the initial vibration signal according to the candidate frequency range corresponding to the first noise; perform correlation analysis on the first time-domain sequence and the second time-domain sequence based on the candidate lag time of the initial vibration signal relative to the initial acoustic signature signal to obtain the analysis result; and, if the analysis result indicates that the first time-domain sequence and the second time-domain sequence are correlated, perform noise reduction processing on the initial vibration signal according to the candidate lag time and the first time-domain sequence to generate a vibration noise-reduced signal.

[0146] In one possible implementation, there are multiple candidate lag times, first time-domain sequences, and / or second time-domain sequences, and the analysis results include multiple correlation values. The first denoising unit 502 is specifically used to: select the maximum correlation value among the multiple correlation values; if the maximum correlation value is greater than a set threshold, determine that the time-domain sequence corresponding to the maximum correlation value in the first time-domain sequence is correlated with the time-domain sequence corresponding to the maximum correlation value in the second time-domain sequence; based on the lag time corresponding to the maximum correlation value in the candidate lag times, shift the time-domain sequence corresponding to the maximum correlation value in the first time-domain sequence in the time domain to obtain a third time-domain sequence; using the third time-domain sequence as a reference signal, perform denoising processing on the initial vibration signal through an adaptive filter to generate a vibration denoising signal.

[0147] In one possible implementation, the second noise reduction unit 503 is specifically used to: perform Fourier transform and resonance feature analysis on the vibration noise reduction signal to determine the mechanical resonance frequency of the vibration noise reduction signal; determine the filter parameters based on the mechanical resonance frequency; and, based on the filter parameters, filter the initial acoustic signature signal through a notch filter or a comb filter to generate an acoustic signature noise reduction signal.

[0148] In one possible implementation, the second noise reduction unit 503 is specifically used to: determine the filter frequency based on the mechanical resonance frequency; and determine the filter bandwidth corresponding to the filter frequency based on the harmonic energy change of the vibration noise reduction signal at the mechanical resonance frequency; wherein the filter parameters include the filter frequency and the filter bandwidth.

[0149] In one possible implementation, the turbine unit valve fault diagnosis device 500 further includes: a third noise reduction unit (not shown in the figure), used to remove environmental noise from the acoustic noise reduction signal by means of an adaptive filter, with the vibration noise reduction signal as a reference signal; and / or a temperature compensation unit (not shown in the figure), used to perform temperature compensation on the vibration amplitude of the vibration noise reduction signal according to temperature data and a pre-built temperature vibration amplitude relationship model.

[0150] In one possible implementation, a pre-trained fault diagnosis model is deployed at the edge. The fault diagnosis unit 504 is specifically used to: acquire the operating parameters of the turbine unit; and perform fault diagnosis on the turbine unit valves based on the vibration noise reduction signal, acoustic noise reduction signal, temperature data and operating parameters, thereby obtaining the fault diagnosis result.

[0151] In one possible implementation, the fault diagnosis unit 504 is specifically used to: input vibration noise reduction signal, acoustic noise reduction signal, temperature data, and operating parameters into the fault diagnosis model; in the fault diagnosis model, based on the operating parameters, analyze the weight parameters corresponding to the vibration noise reduction signal, acoustic noise reduction signal, and temperature data respectively; extract features from the vibration noise reduction signal, acoustic noise reduction signal, and temperature data, and weight the extracted features according to the weight parameters corresponding to the vibration noise reduction signal, acoustic noise reduction signal, and temperature data respectively to obtain a fused feature vector; and perform fault diagnosis on the turbine unit valves based on the fused feature vector to obtain the fault diagnosis result.

[0152] In one possible implementation, the turbine unit valve fault diagnosis device 500 further includes: a sending unit (not shown in the figure) for sending first information to the cloud, the first information indicating vibration noise reduction signal, acoustic noise reduction signal, temperature data and fault diagnosis results; and a receiving unit (not shown in the figure) for receiving second information sent from the cloud, the second information indicating that the acquisition parameters of the sensor device should be updated.

[0153] The turbine unit valve fault diagnosis device 500 provided in this embodiment belongs to the same application concept as the turbine unit valve fault diagnosis method located on one side of the edge provided in the above embodiments of this application. It can execute the turbine unit valve fault diagnosis method located on one side of the edge provided in any of the above embodiments of this application, and has the corresponding functional modules and beneficial effects of executing the turbine unit valve fault diagnosis method located on one side of the edge. Technical details not described in detail in this embodiment can be found in the content of the edge side in the corresponding method embodiments of this application, and will not be repeated here.

[0154] Please see Figure 6 In yet another exemplary embodiment, a turbine generator valve fault diagnosis device 600 is provided. On the cloud side, the turbine generator valve fault diagnosis device 600 includes: a receiving unit 601, a prediction unit 602, a generation unit 603, and a sending unit 604. Wherein:

[0155] The receiving unit 601 is used to receive first information sent by the edge end. The first information indicates vibration noise reduction signal, acoustic noise reduction signal, temperature data, and fault diagnosis result. The vibration noise reduction signal and acoustic noise reduction signal are obtained by collaboratively reducing the initial vibration signal and initial acoustic signal of the turbine unit valve. The fault diagnosis result is obtained by diagnosing the fault of the turbine unit valve based on the vibration noise reduction signal, acoustic noise reduction signal, and temperature data. The prediction unit 602 is used to predict the fault trend of the turbine unit valve based on the vibration noise reduction signal, acoustic noise reduction signal, temperature data, and fault diagnosis result through a fault trend prediction model, and obtain a fault trend prediction result. The generation unit 603 generates second information based on the fault trend prediction result. The second information indicates that the acquisition parameters of the sensor device, the fault diagnosis frequency of the edge end, and / or the data upload frequency of the edge end should be updated. The sending unit 604 is used to send the second information to the edge end.

[0156] The turbine generator valve fault diagnosis device 600 provided in this embodiment belongs to the same concept as the turbine generator valve fault diagnosis method located on the cloud side provided in the above embodiments of this application. It can execute the turbine generator valve fault diagnosis method located on the cloud side provided in any of the above embodiments of this application, and has the corresponding functional modules and beneficial effects for executing the turbine generator valve fault diagnosis method located on the cloud side. Technical details not described in detail in this embodiment can be found in the content on the edge side of the corresponding method embodiment of this application, and will not be repeated here.

[0157] The functions implemented by each unit in the above device can be implemented by the same or different processors, and this application embodiment does not limit this.

[0158] It should be understood that the units in the above device can be implemented by a processor calling software. For example, the device includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of each unit in the device. The processor can be a general-purpose processor, such as a CPU or microprocessor, and the memory can be internal or external to the device. Alternatively, the units in the device can be implemented as hardware circuits. By designing the hardware circuits, some or all of the unit functions can be implemented. The hardware circuits can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are implemented by designing the logical relationships between the components within the circuit. In another implementation, the hardware circuit can be implemented using a PLD, such as an FPGA, which can include a large number of logic gates. The connection relationships between the logic gates are configured through configuration files to implement the functions of some or all of the above units. All units in the above device can be implemented entirely by a processor calling software, entirely by hardware circuits, or partially by a processor calling software with the remaining parts implemented by hardware circuits.

[0159] In this application embodiment, a processor is a circuit with signal processing capabilities. In one implementation, the processor can be a circuit with instruction reading and execution capabilities, such as a CPU, microprocessor, GPU, or DSP. In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits. These logical relationships are fixed or reconfigurable. For example, the processor may be a hardware circuit implemented as an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the processor loading instructions to implement the functions of some or all of the above units. Furthermore, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, or DPU.

[0160] As can be seen, each unit in the above device can be one or more processors (or processing circuits) configured to implement the above methods, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0161] Furthermore, the units in the above devices can be integrated in whole or in part, or they can be implemented independently. In one implementation, these units are integrated together and implemented in the form of a System-on-Chip (SoC). The SoC may include at least one processor for implementing any of the above methods or implementing the functions of the units in the device. The at least one processor may be of different types, such as CPU and FPGA, CPU and artificial intelligence processor, CPU and GPU, etc.

[0162] Exemplary electronic devices

[0163] Another embodiment of this application also proposes an electronic device. See [link to relevant documentation]. Figure 7 As shown, the electronic device may include: a memory 700 and a processor 710; wherein, the memory 700 is connected to the processor 710 and is used to store programs; the processor 710 is used to implement the turbine unit valve fault diagnosis method disclosed in any of the above embodiments by running the programs stored in the memory 700.

[0164] Specifically, the aforementioned electronic device may also include: a bus, a communication interface 720, an input device 730, and an output device 740.

[0165] The processor 710, memory 700, communication interface 720, input device 730, and output device 740 are interconnected via a bus. Among them:

[0166] A bus can include a pathway for transmitting information between various components of a computer system.

[0167] The processor 710 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present application. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0168] The processor 710 may include a main processor, as well as a baseband chip, modem, etc.

[0169] The memory 700 stores a program that executes the technical solution of this application, and may also store an operating system and other critical business functions. Specifically, the program may include program code, which includes computer operation instructions. More specifically, the memory 700 may include read-only memory (ROM), other types of static storage devices capable of storing static information and instructions, random access memory (RAM), other types of dynamic storage devices capable of storing information and instructions, disk storage, flash memory, etc.

[0170] Input device 730 may include a device for receiving data and information input by a user, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.

[0171] Output device 740 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0172] The communication interface 720 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0173] The processor 710 executes the program stored in the memory 700 and calls other devices, which can be used to implement each step of any of the turbine unit valve fault diagnosis methods provided in the above embodiments of this application.

[0174] This application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in the memory through the data interface to execute any of the turbine unit valve fault diagnosis methods provided in the above embodiments. For the specific processing procedure and its beneficial effects, please refer to the above embodiments of the turbine unit valve fault diagnosis method.

[0175] Exemplary computer program products and storage media

[0176] In addition to the methods and devices described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the turbine unit valve fault diagnosis method according to various embodiments of this application as described in any of the above embodiments of this specification.

[0177] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0178] Furthermore, embodiments of this application may also be storage media storing a computer program, which is executed by a processor in the steps of the turbine unit valve fault diagnosis method according to various embodiments of this application described in any of the above embodiments of this specification.

[0179] For the foregoing method embodiments, in order to simplify the description, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0180] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For apparatus embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0181] The steps in the methods of the various embodiments of this application can be adjusted, merged, or deleted in order according to actual needs, and the technical features described in each embodiment can be replaced or combined.

[0182] The modules and sub-modules in the apparatus and terminal in the various embodiments of this application can be merged, divided, and deleted according to actual needs.

[0183] It should be understood that the disclosed terminals, devices, and methods can be implemented in other ways, given the several embodiments provided in this application. For example, the terminal embodiments described above are merely illustrative. For instance, the division of modules or sub-modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple sub-modules or modules may be combined or integrated into another module, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0184] The modules or submodules described as separate components may or may not be physically separate. The components that constitute a module or submodule may or may not be physical modules or submodules; that is, they may be located in one place or distributed across multiple network modules or submodules. Some or all of the modules or submodules can be selected to achieve the purpose of this embodiment's solution, depending on actual needs.

[0185] Furthermore, the functional modules or sub-modules in the various embodiments of this application can be integrated into one processing module, or each module or sub-module can exist physically separately, or two or more modules or sub-modules can be integrated into one module. The integrated modules or sub-modules described above can be implemented in hardware or in the form of software functional modules or sub-modules.

[0186] 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 implementation should not be considered beyond the scope of this application.

[0187] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software unit executed by a processor, or a combination of both. The software unit can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0188] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0189] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for diagnosing valve faults in a hydroelectric turbine unit, characterized in that, Applied to the edge end deployed on one side of the turbine unit, including: Data is collected from the turbine unit valves using sensor equipment to obtain initial acoustic signature signals, initial vibration signals, and temperature data. Using the initial acoustic signature signal as a reference signal, and based on the time-domain correlation between the fluid noise in the initial acoustic signature signal and the first noise in the initial vibration signal, the initial vibration signal is denoised to generate a vibration denoising signal. The first noise is the vibration noise caused by the transmission of the fluid noise in the valve body structure. Using the vibration noise reduction signal as a reference signal, and based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial acoustic pattern signal, the initial acoustic pattern signal is denoised to generate an acoustic pattern noise reduction signal. The second noise is the noise transmitted from the mechanical resonance noise to the acoustic sensor through the valve body structure. Based on the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data, fault diagnosis is performed on the turbine unit valves to obtain fault diagnosis results.

2. The method for diagnosing valve faults in a hydro-turbine unit according to claim 1, characterized in that, The step of using the initial acoustic signature signal as a reference signal, and performing noise reduction processing on the initial vibration signal based on the characteristic correlation between the fluid noise in the initial acoustic signature signal and the first noise in the initial vibration signal to generate a vibration noise-reduced signal includes: Based on the candidate frequency range corresponding to the fluid noise, a first time-domain sequence is obtained from the initial acoustic signature signal; Based on the candidate frequency range corresponding to the first noise, a second time-domain sequence is obtained from the initial vibration signal; Based on the candidate lag time of the initial vibration signal relative to the initial acoustic signature signal, a correlation analysis is performed on the first time-domain sequence and the second time-domain sequence to obtain the analysis results; If the analysis results indicate that the first time-domain sequence is related to the second time-domain sequence, the initial vibration signal is denoised based on the candidate lag time and the first time-domain sequence to generate the denoised vibration signal.

3. The method for diagnosing valve faults in a hydro-turbine unit according to claim 2, characterized in that, The number of candidate lag times, the first time-domain sequence, and / or the second time-domain sequence is multiple, and the analysis results include multiple correlation values; When the analysis results indicate that the first time-domain sequence is correlated with the second time-domain sequence, the initial vibration signal is denoised based on the candidate lag time and the first time-domain sequence to generate the denoised vibration signal, including: Among the multiple relevant values, the largest relevant value is selected; If the maximum correlation value is greater than a set threshold, it is determined that the time-domain sequence corresponding to the maximum correlation value in the first time-domain sequence is correlated with the time-domain sequence corresponding to the maximum correlation value in the second time-domain sequence. Based on the lag time corresponding to the maximum correlation value among the candidate lag times, the time domain sequence corresponding to the maximum correlation value in the first time domain sequence is shifted in the time domain to obtain the third time domain sequence; Using the third time-domain sequence as a reference signal, the initial vibration signal is denoised using an adaptive filter to generate the denoised vibration signal.

4. The method for diagnosing valve faults in a hydro-turbine unit according to claim 1, characterized in that, The step of using the vibration noise reduction signal as a reference signal, and based on the frequency domain correlation between the mechanical resonance noise in the vibration noise reduction signal and the second noise in the initial acoustic signature signal, performing noise reduction processing on the initial acoustic signature signal to generate an acoustic signature noise reduction signal includes: Fourier transform and resonance characteristic analysis are performed on the vibration noise reduction signal to determine the mechanical resonance frequency of the vibration noise reduction signal; Determine the filter parameters based on the mechanical resonance frequency; Based on the filter parameters, the initial voiceprint signal is filtered using a notch filter or a comb filter to generate the voiceprint noise reduction signal.

5. The method for diagnosing valve faults in a hydro-turbine unit according to claim 4, characterized in that, Determining the filter parameters based on the mechanical resonant frequency includes: The filter frequency is determined based on the mechanical resonance frequency. The filter bandwidth corresponding to the filter frequency is determined based on the harmonic energy change of the vibration noise reduction signal at the mechanical resonance frequency. The filter parameters include the filter frequency and the filter bandwidth.

6. The method for diagnosing valve faults in a hydro-turbine unit according to any one of claims 1 to 5, characterized in that, Before obtaining the fault diagnosis result by performing fault diagnosis on the turbine unit valves based on the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data, the following operations are also included: Using the vibration noise reduction signal as a reference signal, an adaptive filter is used to remove environmental noise from the acoustic noise reduction signal; And / or, based on the temperature data and a pre-built temperature-vibration amplitude relationship model, temperature compensation is applied to the vibration amplitude of the vibration noise reduction signal.

7. The method for diagnosing valve faults in a hydro-turbine unit according to any one of claims 1 to 5, characterized in that, The pre-trained fault diagnosis model is deployed at the edge, and fault diagnosis is performed on the turbine unit valves based on the vibration noise reduction signal, the acoustic noise reduction signal, and the temperature data to obtain fault diagnosis results, including: Obtain the operating parameters of the turbine unit; Based on the vibration noise reduction signal, the acoustic noise reduction signal, the temperature data, and the operating parameters, the fault diagnosis model is used to diagnose the faults of the turbine unit valves, and the fault diagnosis results are obtained.

8. A valve fault diagnosis system for a hydro-turbine unit, characterized in that, include: The edge end is deployed on one side of the turbine unit; The edge end is configured to perform the turbine unit valve fault diagnosis method as described in any one of claims 1 to 7.

9. An electronic device, characterized in that, Including memory and processor; The memory is connected to the processor and is used to store programs; The processor is used to implement the turbine unit valve fault diagnosis method as described in any one of claims 1 to 7 by running the program in the memory.

10. A computer program product, characterized in that, The system includes a computer program that, when executed by a processor, implements the turbine unit valve fault diagnosis method as described in any one of claims 1 to 7.

Images