A method and system for monitoring the health status of a turbine guide vane

Abstract

The present invention provides a method and system for monitoring the health of turbine guide vanes, relating to the technical field of turbine health monitoring. The method comprises: using an acoustic wave sensor and a vibration sensor installed at the turbine's volute manhole door to respectively collect acoustic and vibration signals generated during guide vane operation; extracting the time domain features, frequency domain features, and time-frequency combined features of the acoustic and vibration signals; inputting the time domain features, frequency domain features, and time-frequency combined features into a pre-trained fault diagnosis model, and monitoring the health of the guide vanes using the fault diagnosis model. The present invention can monitor the operating status of the guide vanes in real time, accurately determine their health status, and promptly detect abnormalities such as water leaks, thereby providing accurate maintenance recommendations.

Description

Technical Field

[0001] The present invention relates to the technical field of water turbine health monitoring, and in particular to a water turbine guide vane health status monitoring method and system. Background Art

[0002] As the core equipment for hydroelectric power generation, the health of a turbine's guide vanes is directly related to power generation efficiency and safety. Damage to the guide vanes can easily lead to water leakage, affecting power generation efficiency. During daily operation, when the unit is in operation or standby mode, the runner chamber is fully filled with water, making it difficult to inspect the health of the guide vanes. Furthermore, when the turbine is operating normally, the water flow in the runner chamber is high, making it impossible to install precision sensors to monitor the health of the guide vanes.

[0003] Traditional methods for determining the health of guide vanes rely primarily on manual judgment and radiographic testing after drainage and overhaul. Manual judgment refers to production personnel applying their personal experience in power production to determine the health of guide vanes. For example, by combining changes in the unit's water consumption rate, they can infer leaks and determine that the guide vanes are unhealthy. Radiographic testing after drainage and overhaul refers to the process of draining the water after overhaul, allowing personnel to enter the volute and carry out inspections of the guide vanes using relevant equipment.

[0004] The traditional method of judging the health of guide vanes has the following problems:

[0005] 1. Guide vane sealing performance degradation and structural damage are difficult to detect in a timely manner. Using manual experience to determine the damage is overly dependent on the responsibility and personal ability of production personnel. Radiographic inspection of the guide vanes after drainage maintenance requires that the unit be inspected during maintenance.

[0006] 2. Abnormal water flow and water pressure changes are difficult to identify accurately, and abnormal conditions such as guide vane leakage cannot be discovered in time, resulting in increased water energy loss.

[0007] 3. Guide vane abnormalities are difficult to repair in a timely manner. Due to the lack of effective guide vane health monitoring methods, guide vane abnormalities are often difficult to detect and accurately judge in a timely manner. Regular maintenance after complete drainage is required (maintenance intervals can be as long as six months to one year). As a result, guide vane leakage hazards cannot be discovered and repaired in a timely manner, affecting the operating efficiency and safety of the unit.

[0008] 4. Inadequate guide vane health monitoring system performance. Existing guide vane health monitoring methods often cannot accurately analyze the overall health of the guide vanes, nor can they accurately predict potential guide vane failures and maintenance needs, making preventive maintenance difficult.

[0009] Therefore, traditional turbine guide vane health monitoring methods have many shortcomings and cannot meet the needs of modern hydropower generation for guide vane health monitoring. Based on this, a safe, reliable and accurate turbine guide vane health status monitoring method is urgently needed to determine the health of the guide vanes. Summary of the Invention

[0010] The purpose of the present invention is to provide a method and system for monitoring the health status of turbine guide vanes, which can solve the problems that traditional methods for monitoring the health of turbine guide vanes have many shortcomings and are difficult to meet the needs of modern hydropower generation for guide vane health monitoring. The system can monitor the operating status of the guide vanes in real time, accurately analyze the health status of the guide vanes, and promptly detect abnormal conditions such as water leakage, thereby providing accurate maintenance suggestions.

[0011] To achieve the above objectives, in a first aspect, the present invention provides a method for monitoring the health status of a turbine guide vane, comprising:

[0012] The acoustic wave sensor and vibration sensor installed at the manhole door of the turbine volute respectively collect the acoustic wave signal and vibration signal generated by the guide vane during operation;

[0013] Extract the time domain features, frequency domain features and time-frequency joint features of the acoustic and vibration signals;

[0014] The time domain features, frequency domain features and time-frequency joint features are input into the pre-trained fault diagnosis model, and the health status of the guide vanes is monitored through the fault diagnosis model.

[0015] According to a method for monitoring the health status of a turbine guide vane provided by the present invention, time domain features include kurtosis, form factor and zero-crossing rate, frequency domain features include frequency band energy ratio, and time-frequency joint features include the cross-correlation peak delay between the acoustic signal and the vibration signal.

[0016] According to a method for monitoring the health status of a turbine guide vane provided by the present invention, the kurtosis is calculated as follows:

[0017]

[0018]

[0019] Where, is the steepness, Represents the fourth-order central moment of the acoustic signal; Indicates the standard deviation of the sound wave signal; Represents the mean value of the sound wave signal; N Indicates the number of sampling points in a single frame of the sound wave signal; Represents the sound wave signal at the i-th sampling point.

[0020] According to a method for monitoring the health status of a turbine guide vane provided by the present invention, the calculation formula of the form factor is:

[0021]

[0022] Where, C is the form factor; Indicates the i Vibration signal of sampling points; M Indicates the number of sampling points in a single frame of the vibration signal.

[0023] According to a method for monitoring the health status of a turbine guide vane provided by the present invention, the calculation formula for the zero-crossing rate is:

[0024]

[0025] Where, is the zero-crossing rate; Indicates the i Vibration signal of sampling points; Indicates the i +1 sampling point vibration signal; M Indicates the number of sampling points in a single frame of vibration signal; i The sampling point and i When the signs of the vibration signals at +1 sampling points are opposite, ; When the i The sampling point and i When the signs of the vibration signals of +1 sampling points are the same, .

[0026] According to a method for monitoring the health status of a turbine guide vane provided by the present invention, the frequency band energy ratio is calculated as follows:

[0027]

[0028] Where, is the frequency band energy ratio; is the first frequency components, Indicates the lowest frequency, Indicates the highest frequency; N Indicates the number of sampling points in a single frame of the sound wave signal.

[0029] According to a method for monitoring the health of a turbine guide vane provided by the present invention, the cross-correlation peak delay between the acoustic signal and the vibration signal is:

[0030]

[0031]

[0032] Where, Indicates the cross-correlation peak delay between the acoustic signal and the vibration signal; Represents the sound wave signal, Indicates vibration signal, Indicates time delay, Indicates time delay Correlation between acoustic wave signal and vibration signal.

[0033] According to a method for monitoring the health of a turbine guide vane provided by the present invention, the training process of a pre-trained fault diagnosis model includes:

[0034] Collect multiple sets of test data of the guide vanes operating in healthy conditions and fault conditions; each set of test data includes acoustic and vibration signals;

[0035] Extract historical time domain features, historical frequency domain features and historical time-frequency joint features from the acoustic signal and vibration signal of each set of test data respectively;

[0036] The historical time domain features, historical frequency domain features, and historical time-frequency joint features corresponding to each set of test data are taken as a sample, and the guide vane state corresponding to the set of test data is used as the label of the sample to construct a training sample;

[0037] The initial model is trained using multiple training samples consisting of multiple groups of test data to obtain a pre-trained fault diagnosis model.

[0038] According to a method for monitoring the health status of a turbine guide vane provided by the present invention, the initial model is a support vector machine.

[0039] In a second aspect, the present invention provides a turbine guide vane health status monitoring system, comprising:

[0040] A collection unit is used to collect the acoustic wave signal and vibration signal generated by the guide vane during operation through an acoustic wave sensor and a vibration sensor installed at the volute manhole door of the turbine;

[0041] An extraction unit, used to extract time domain features, frequency domain features and time-frequency joint features of sound wave signals and vibration signals;

[0042] The monitoring unit is used to input the time domain features, frequency domain features and time-frequency joint features into a pre-trained fault diagnosis model, and monitor the health status of the guide vanes through the fault diagnosis model.

[0043] The technical solution of the present invention has at least the following technical effects:

[0044] The present invention provides a method and system for monitoring the health of turbine guide vanes. The method comprises: using an acoustic sensor and a vibration sensor installed at the turbine's volute manhole door to respectively collect acoustic and vibration signals generated by the guide vanes during operation; extracting the time-domain, frequency-domain, and time-frequency combined features of the acoustic and vibration signals; inputting the time-domain, frequency-domain, and time-frequency combined features into a pre-trained fault diagnosis model, and monitoring the health of the guide vanes using the fault diagnosis model. By utilizing a non-invasive sensor layout design and collecting multimodal signal features, the system can monitor the operating status of the guide vanes in real time, analyze their health, and promptly detect abnormalities such as leaks, thereby providing maintenance recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] In the attached figure:

[0047] Figure 1 The figure is a flow chart of the method for monitoring the health status of a turbine guide vane according to the present invention. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0049] The following will describe some embodiments of the present invention in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0050] See also Figure 1 The embodiment of the present invention provides a method for monitoring the health status of a turbine guide vane, comprising:

[0051] Step 1: Using an acoustic wave sensor and a vibration sensor installed at the volute manhole door of the turbine, respectively collect acoustic wave signals and vibration signals generated when the guide vanes are running;

[0052] It should be noted that during power generation, a large amount of water flows into the runner chamber, resulting in extremely high water volume and pressure. Therefore, sensors cannot be installed near the guide vanes inside the runner chamber. Sensors and other monitoring devices must be installed in a non-moving area, such as outside the runner chamber. The volute manhole door is located outside the volute, where there is no moving water. Therefore, the present invention installs acoustic and vibration sensors there, protecting them from the impact of the water flow.

[0053] Specifically, the acoustic sensor has a sampling rate of 10kHz and a range of 30-130dB. The vibration sensor can be a triaxial accelerometer with a range of ±10g (g is the acceleration due to gravity) and a sampling rate of 2kHz. To prevent direct water flow, the acoustic and vibration sensors can be installed outside the volute manhole door.

[0054] In some embodiments, the collected sound wave signals and vibration signals may be subjected to wavelet denoising and time-frequency alignment to eliminate the time deviation between the ambient noise and the sound wave sensor and the vibration sensor.

[0055] Step 2: Extract the time domain features, frequency domain features and time-frequency joint features of the sound wave signal and the vibration signal;

[0056] Specifically, the time domain features include kurtosis, shape factor and zero-crossing rate.

[0057] The calculation formula of kurtosis is:

[0058]

[0059]

[0060] Where, is the steepness, Represents the fourth-order central moment of the acoustic signal; Indicates the standard deviation of the sound wave signal; Represents the mean value of the sound wave signal; N Indicates the number of sampling points in a single frame of the sound wave signal; Represents the sound wave signal at the i-th sampling point.

[0061] Kurtosis is suitable for detecting transient shocks or abnormal pulses in acoustic signals. Under normal operating conditions, the kurtosis of a guide vane acoustic signal is low. However, leakage caused by a guide vane failure will gradually increase the kurtosis. In a minor fault, the acoustic signal will have a small number of transient shocks, and the kurtosis will gradually increase. In a severe fault, the acoustic signal will have frequent transient shocks with a sharp distribution, and the kurtosis will reach its maximum value. Therefore, the physical significance of kurtosis is to detect transient shocks caused by guide vane damage and leakage.

[0062] The calculation formula for the form factor is:

[0063]

[0064] Where, C The form factor is an important time domain feature that measures the relationship between the peak intensity and overall energy of the vibration signal. It is defined as the ratio of the peak value to the effective value of the vibration signal. Indicates the i Vibration signal of sampling points; M Represents the number of sampling points within a single frame of the vibration signal. Under normal operating conditions, the waveform of the guide vane vibration signal is relatively smooth, with a small difference between the peak value and the effective value, and a small form factor. When a guide vane malfunctions, causing leakage or other phenomena, the mechanical structure suddenly releases energy, generating a transient, high-amplitude impact signal (collision vibration, water leakage vibration, etc.). At this time, the peak value increases significantly, while the effective value grows more slowly due to the time averaging effect, causing the form factor to increase dramatically. For example, the form factor of a flat square wave can be as low as 1.0 (unit form factor), but for other more peaky waveforms (such as triangle waves), the form factor can be as high as 3 or 4. When a guide vane malfunctions, the form factor increases suddenly.

[0065] The calculation formula for the zero-crossing rate is:

[0066]

[0067] Where, The zero-crossing rate indicates the number of times the vibration signal crosses the zero axis per unit time. Indicates the i Vibration signal of sampling points; M Indicates the number of sampling points in a single frame of vibration signal; It is 1 when the signs of the vibration signals of adjacent sampling points are opposite, otherwise it is 0. Under normal working conditions, the water flow and mechanical vibration signals are relatively stable, and the zero crossing rate Small and stable; when the guide vane leaks under fault conditions, the vibration signal oscillates rapidly and the zero crossing rate Significantly increased.

[0068] Specifically, the frequency domain features include frequency band energy ratios, such as the energy ratios of the vibration signal in the 0-2000 Hz frequency band.

[0069] The calculation formula of the frequency band energy ratio is:

[0070]

[0071] Where, is the first frequency components, Represents the lowest frequency, Indicates the highest frequency; Nis the acoustic signal frame length, representing the number of sampling points within a single acoustic signal frame. The numerator is the sum of the energies of all frequency components within the target frequency band, while the denominator is the total energy of the signal across the entire frequency band (0 minus the Nyquist frequency, which is half the system sampling frequency).

[0072] The frequency band energy ratio (FHR) is the acoustic signal's frequency band energy ratio, calculating the ratio of the energy in a specific frequency range (frequency band) to the total signal energy. This metric quantifies the contribution of different frequency bands to the signal and is a key characteristic for distinguishing normal from faulty conditions in guide vane health monitoring. Historical data shows that the dominant frequency of water leakage cavitation noise is typically at higher frequencies (500 Hz to 2 kHz). Guide vane faults stimulate acoustic energy in specific frequency bands. Therefore, measuring the proportion of high-frequency energy under current operating conditions can distinguish between normal and faulty conditions. This method is also resistant to noise interference, as environmental noise (such as water turbulence) is typically distributed across the entire frequency range (0 to 2 kHz), while fault energy is concentrated in a specific frequency band. By calculating the energy ratio, fault signatures can be amplified and noise influences can be suppressed. Under normal conditions, the frequency band energy of a guide vane is relatively low, but increases significantly when a guide vane fault occurs.

[0073] Specifically, the time-frequency joint feature includes the cross-correlation peak delay between the acoustic signal and the vibration signal.

[0074] The cross-correlation function measures the similarity of two signals at different time delays. The peak delay finds the time difference that makes the two signals most similar. Under normal circumstances, vibration signals propagate along a fixed structural path. When a guide vane fails, the vibration wave propagates along an alternative path, causing the time delay of the vibration signal relative to the acoustic signal to change. By calculating the change in the cross-correlation peak delay between the acoustic and vibration signals, changes in the mechanical state of the guide vane (such as looseness and wear) can be effectively detected, providing key timing characteristics for fault diagnosis.

[0075] The mathematical expression of the cross-correlation function is:

[0076]

[0077] in, It represents the acoustic wave signal (this signal is the reference signal). It represents the vibration signal (signal to be aligned). It represents the time delay (integer multiple of the sampling interval), Indicates the time delay When The maximum value indicates that the two signals are most similar at this time delay, that is, the time delay is the cross-correlation peak delay of the vibration signal relative to the acoustic signal. When the guide vane condition changes from healthy to unhealthy, the cross-correlation peak delay is Mutations will occur.

[0078] Therefore, the cross-correlation peak delay between the acoustic signal and the vibration signal is:

[0079]

[0080] in, The meaning of the representative is hour, Take the maximum value, The maximum value indicates that the two signals are most similar under this delay, that is, the delay is the propagation delay of the vibration signal relative to the acoustic signal. Sudden changes may occur. For example, when the guide vane is damaged and leaks Shorten (for example, from the normal value of 5ms to 2ms).

[0081] Step 3: Input the time domain features, frequency domain features, and time-frequency joint features into a pre-trained fault diagnosis model, and monitor the health status of the guide vanes through the fault diagnosis model.

[0082] Specifically, the training process of the pre-trained fault diagnosis model includes:

[0083] Collect multiple sets of test data of the guide vanes operating in healthy conditions and fault conditions; each set of test data includes acoustic and vibration signals;

[0084] Extract historical time domain features, historical frequency domain features and historical time-frequency joint features from the acoustic signal and vibration signal of each set of test data respectively;

[0085] The historical time domain features, historical frequency domain features, and historical time-frequency joint features corresponding to each set of test data are taken as a sample, and the guide vane state corresponding to the set of test data is used as the label of the sample to construct a training sample;

[0086] The initial model is trained using multiple training samples consisting of multiple groups of test data to obtain a pre-trained fault diagnosis model.

[0087] In some embodiments, the initial model may be a support vector machine.

[0088] It's important to note that the Support Vector Machine (SVM) is a commonly used machine learning algorithm, particularly well-suited for small-sample, nonlinear binary classification problems. Its basic concept is to find an optimal classification hyperplane in the feature space that separates samples of different categories with the greatest possible margin. For device status recognition, healthy and faulty states can be considered two categories, and the multidimensional features collected from the device are considered as sample points. Using the SVM, an optimal state discrimination model can be constructed in the feature space.

[0089] The SVM model is trained using historical data from the healthy sample library. The main steps are as follows: a certain number of samples are selected from the healthy sample library as the training set. Each sample contains multiple feature variables and corresponding status labels. The healthy status label is -1 and the fault status label is +1. All feature variables are normalized so that their numerical range is scaled to [0, 1] or [-1, +1] to avoid the influence of features with different dimensions.

[0090] For the linearly separable case, the goal of the SVM is to find an optimal hyperplane that separates the two classes of samples by the hyperplane with the maximum margin. For the linearly inseparable case, slack variables and penalty factors can be introduced to allow for a small number of sample misclassifications while controlling the cost of misclassification through the penalty factor. This problem can be optimally solved by solving a convex quadratic programming problem. The optimal hyperplane parameters obtained can be used to construct the decision function of the SVM. For nonlinear classification problems, the kernel technique can be used to map samples from the original space to a high-dimensional feature space, making them linearly separable.

[0091] Common kernel functions include polynomial kernels and Gaussian kernels (RBF kernels). Using training data, the optimal model parameters are obtained by solving a convex quadratic programming problem for the Support Vector Machine (SVM). For nonlinear SVMs, the kernel function parameters must be carefully selected. A common approach is to combine cross-validation with grid search to select the parameter combination with the best average performance. During training, a balance must be struck between the model's fitting ability and generalization performance. By adjusting the penalty factor or kernel function parameters, the model's complexity can be controlled to avoid overfitting.

[0092] Once the SVM model is trained, it can be used to perform real-time diagnosis of the guide vane health status using online monitoring data. Multiple feature quantities are extracted from the collected vibration and acoustic signals in the same manner as in the training phase. The feature quantities can be normalized to align with the numerical range of the training samples. The pre-processed feature vector is substituted into the decision function of the SVM model to obtain the label of the guide vane's current state. (+1 or -1). Combined with the operating conditions of the guide vanes, further interpret the status tag: If If it is -1, the guide vane is in normal operation; if If is +1, the guide vane may be faulty or have performance degradation. Of course, for the guide vane monitoring data identified as faulty or abnormal, its feature vector can be compared with the support vector of the SVM to find the most similar fault mode and preliminarily infer the fault type and cause.

[0093] In practical applications, the fault diagnosis model trained using this SVM model can be embedded into a turbine online monitoring system (e.g., an embedded industrial computer) to perform real-time diagnosis on newly acquired vibration and acoustic signals. If the guide vane status is predicted to be abnormal five times in a row, a system alert is triggered.

[0094] Based on the same inventive concept, another embodiment of the present invention provides a turbine guide vane health status monitoring system. The system corresponds to the method of the aforementioned embodiment and includes:

[0095] A collection unit is used to collect the acoustic wave signal and vibration signal generated by the guide vane during operation through an acoustic wave sensor and a vibration sensor installed at the volute manhole door of the turbine;

[0096] An extraction unit, used to extract time domain features, frequency domain features and time-frequency joint features of sound wave signals and vibration signals;

[0097] The monitoring unit is used to input the time domain features, frequency domain features and time-frequency joint features into a pre-trained fault diagnosis model, and monitor the health status of the guide vanes through the fault diagnosis model.

[0098] The following is a specific embodiment of the present invention.

[0099] The experimental environment was located in the turbine governor area of ​​a hydropower station. The turbine was an axial-flow propeller type with a rated head of 32.5 m and a flow rate of 0 m³ / s (under normal shutdown conditions). The acoustic sensor had a sampling rate of 10 kHz and a range of 30 to 130 dB. The vibration sensor used a triaxial accelerometer (X / Y / Z) with a sampling rate of 2 kHz and a range of ±10 g.

[0100] The guide vanes are running in a healthy state (normal working condition). At this moment, the guide vanes are fully closed. The acoustic sensor and vibration sensor are running for 2 hours, collecting acoustic signals and vibration signals respectively. .

[0101] The guide vane operates in a faulty state (faulty operating condition). The guide vane leakage fault is artificially simulated, and the guide vane opening is expanded to 2mm (leakage, at this time the guide vane opening is about 1%). The acoustic wave sensor and vibration sensor are operated continuously for 1 hour to collect acoustic wave signals and vibration signals respectively. .

[0102] Multiple sets of data set samples of guide vanes are obtained, including: training set: 800 normal samples, 200 fault samples (at a ratio of 4:1); test set: 200 normal samples, 50 fault samples.

[0103] Preprocess the acoustic wave signal and vibration signal, including:

[0104] Denoising:

[0105] The acoustic signal is decomposed into 5 layers using the Sym5 wavelet basis, and the SURE algorithm is used for adaptive threshold denoising.

[0106] The vibration signal is decomposed using a 4-layer wavelet and soft threshold processing, retaining the 0~2000Hz frequency band.

[0107] Framing and standardization:

[0108] Framing parameters: 0.1 second frame length (1000 points for acoustic signals, 200 points for vibration signals), 50% overlap. Z-score normalization: mean and standard deviation calculated per frame.

[0109] Frame calculation is performed every 0.1 seconds, and the following five features are extracted from each frame signal:

[0110] 1. Time domain characteristics:

[0111] Kurtosis : Normally around 3.0, it rises to 8.5 or above in the event of a fault.

[0112] Form Factor : Normally around 3.1, it rises to 5.2 or above in the event of a fault.

[0113] Zero-crossing rate : 150 times / second in normal operation, rising to 300 or above in case of failure.

[0114] 2. Frequency domain characteristics:

[0115] Vibration signal 0~2000Hz frequency band energy ratio : The energy of the 500-2000HZ frequency band is 5% in normal state and can reach 25% in case of fault.

[0116] 3. Joint time-frequency features:

[0117] Peak delay of cross-correlation between acoustic and vibration signals : 5ms under normal conditions, which can be shortened to 2ms in case of failure.

[0118] The prediction results of the present invention are:

[0119]

[0120] This example uses an SVM-based fault diagnosis model combined with the aforementioned features to achieve high-precision detection of guide vane water leakage faults (recall rate of 90.5% and F1 score of 92.3%), verifying the following advantages:

[0121] Category imbalance processing: can increase the recall rate of fault classes by 30%.

[0122] Real-time: Single prediction time is less than 1ms, meeting the millisecond-level response requirements of hydropower stations.

[0123] Interpretability: Features such as cross-correlation delay and kurtosis are key features that guide optimal sensor deployment.

[0124] This solution can be directly integrated into the existing monitoring system of the hydropower station without modifying the guide vane structure, and uses non-invasive sensors to provide reliable support for guide vane health management.

[0125] Compared with the prior art, the beneficial effects of the technical solution of the present invention are as follows:

[0126] 1. Improve monitoring accuracy: Through the technical means of vibration sensors and acoustic wave sensors, multiple features, and machine learning algorithms, the present invention can more accurately monitor the operating status of the guide vanes and promptly detect abnormal conditions such as water leakage.

[0127] 2. Reduce power generation losses: The present invention can detect the phenomenon of guide vane leakage expansion as early as possible, carry out hidden danger inspection in time, and reduce power generation losses caused by leakage.

[0128] 3. Improve the operating efficiency and safety of the unit: By promptly detecting abnormal conditions of the guide vanes and providing maintenance suggestions, the present invention helps to improve the operating efficiency and safety of the unit.

[0129] 4. Reduce maintenance costs and risks: The present invention can accurately analyze the health of the guide vanes and provide accurate maintenance suggestions, thereby reducing maintenance costs and risks.

[0130] 5. Implement preventive maintenance: Through quantitative data analysis, the present invention can predict potential failures and maintenance needs, implement preventive maintenance, and reduce the impact of sudden failures on power generation.

[0131] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the embodiments disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and variations can be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for monitoring the health status of a turbine guide vane, characterized in that: The method is used for detecting a water leakage fault of a turbine guide vane, and the method comprises: The acoustic wave sensor and vibration sensor installed at the manhole door of the turbine volute respectively collect the acoustic wave signal and vibration signal generated by the guide vane during operation; Extracting time domain features, frequency domain features, and time-frequency joint features of the sound wave signal and the vibration signal; Inputting the time domain features, frequency domain features and time-frequency joint features into a pre-trained fault diagnosis model, and monitoring the health status of the guide vanes through the fault diagnosis model; The time domain features include kurtosis, form factor and zero-crossing rate, the frequency domain features include frequency band energy ratio, and the time-frequency joint features include the cross-correlation peak delay between the acoustic signal and the vibration signal; The calculation formula of the kurtosis is: Where K is the kurtosis, Represents the fourth-order central moment of the acoustic signal; Indicates the standard deviation of the sound wave signal; Represents the mean value of the sound wave signal; N Indicates the number of sampling points in a single frame of the sound wave signal; Represents the acoustic wave signal of the i-th sampling point; The calculation formula of the shape factor is: Where, C is the form factor; Indicates the i Vibration signal of sampling points; M Indicates the number of sampling points in a single frame of vibration signal; The calculation formula of the zero-crossing rate is: Where Z is the zero-crossing rate; Indicates the i Vibration signal of sampling points; Indicates the i +1 sampling point vibration signal; M Indicates the number of sampling points in a single frame of vibration signal; i The sampling point and i When the signs of the vibration signals at +1 sampling points are opposite, ; When the i The sampling point and i When the signs of the vibration signals of +1 sampling points are the same, ; The calculation formula of the frequency band energy ratio is: Where, is the frequency band energy ratio; is the kth frequency component of the sound wave signal after fast Fourier transform, Indicates the lowest frequency, Indicates the highest frequency; N Indicates the number of sampling points in a single frame of the sound wave signal; The cross-correlation peak delay between the acoustic signal and the vibration signal is: Where, Indicates the cross-correlation peak delay between the acoustic signal and the vibration signal; Represents the sound wave signal, Indicates vibration signal, Indicates time delay, Indicates time delay Correlation between acoustic wave signal and vibration signal.

2. The method for monitoring the health status of a turbine guide vane according to claim 1, wherein: The training process of the pre-trained fault diagnosis model includes: Collect multiple sets of test data of the guide vanes operating in healthy conditions and fault conditions; each set of test data includes acoustic and vibration signals; Extract historical time domain features, historical frequency domain features and historical time-frequency joint features from the acoustic signal and vibration signal of each set of test data respectively; The historical time domain features, historical frequency domain features, and historical time-frequency joint features corresponding to each set of test data are taken as a sample, and the guide vane state corresponding to the set of test data is used as the label of the sample to construct a training sample; The initial model is trained using multiple training samples consisting of multiple groups of test data to obtain a pre-trained fault diagnosis model.

3. The method for monitoring the health status of a turbine guide vane according to claim 2, wherein: The initial model is a support vector machine.

4. A turbine guide vane health status monitoring system, characterized in that: The method for monitoring the health status of a turbine guide vane according to claim 1 is adopted, wherein the system comprises: A collection unit is used to collect the acoustic wave signal and vibration signal generated by the guide vane during operation through an acoustic wave sensor and a vibration sensor installed at the volute manhole door of the turbine; An extraction unit, configured to extract time domain features, frequency domain features, and time-frequency joint features of the sound wave signal and the vibration signal; The monitoring unit is used to input the time domain features, frequency domain features and time-frequency joint features into a pre-trained fault diagnosis model, and monitor the health status of the guide vanes through the fault diagnosis model.

Images