Water turbine cavitation prediction method and system based on support vector machine

Abstract

The invention relates to the technical field of water turbine monitoring, in particular to a water turbine cavitation prediction method and system based on a support vector machine, which directly focuses on the physical nature of an internal flow field causing cavitation through an advanced means of numerical simulation, establishes a mapping model of a corresponding cavitation risk through the support vector machine, and performs cavitation prediction on the internal flow field. And real-time accurate early warning of the cavitation state is realized. According to the method, a full-parametric modeling and corrected physical model is adopted, and the method has the remarkable advantages of being low in cost, high in reproducibility and easy to popularize in water turbines of different models and working conditions, so that a prospective technical guarantee is provided for safe, stable and efficient operation of the water turbines.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of water turbine monitoring, in particular to a water turbine cavitation prediction method and system based on a support vector machine. BACKGROUND

[0002] As the core equipment for converting water energy, the operation stability of a water turbine is directly related to the benefit and safety of a hydropower station. During operation, when the local pressure in the flow passage is lower than the saturated steam pressure of the water at the temperature, cavitation occurs, leading to the growth and collapse of gas nuclei in the fluid. Long-term effects will cause cavitation erosion damage to the flow components such as runner blades, not only causing the unit efficiency to decrease and the output to fluctuate, but also causing abnormal vibration and noise of the unit, seriously threatening the safety and service life of the unit, and causing huge economic losses. Therefore, effective monitoring and early warning of water turbine cavitation is a key technical problem that the hydropower industry has been committed to solving for a long time.

[0003] To address the technical challenge of water turbine cavitation monitoring and diagnosis, existing technologies have proposed various solutions, but all have corresponding limitations. The first type of method is based on semi-physical simulation and sensor data acquisition, through three-dimensional scanning modeling and relying on field sensor data for model correction. Although it has certain prediction ability, it has the defects of high modeling cost, long cycle and poor universality. The second type of method widely uses signal processing and machine learning technology, through the acquisition of physical signals such as acoustic emission, vibration, pressure pulsation, combined with modal decomposition, entropy feature extraction and classification algorithm to realize cavitation recognition. Although it improves the accuracy of state recognition, it is essentially a post-diagnosis and cannot capture the critical state of cavitation inception, with lagging early warning and high computational complexity, limiting real-time performance. The third type of method is based on image recognition principles, through visual detection of cavitation bubbles, but is limited by observation conditions, water transparency and environmental light, and has insufficient stability in engineering applications.

[0004] In summary, the existing technical solutions mainly face the following three levels of limitations in response to cavitation early warning needs: First, in terms of technology, they rely heavily on external indirect signals such as vibration and noise, failing to directly link to the physical nature of the flow field at the inception of cavitation, resulting in delayed recognition and insufficient early warning capabilities. Secondly, in terms of implementation methods, many solutions rely heavily on physical sensor networks or specific hardware systems, resulting in complex deployment, high economic costs, and signal acquisition quality being easily disturbed by the on-site environment, making it difficult to migrate and reproduce in different units and operating conditions. Thirdly, in terms of system functions, existing technologies focus on state recognition and degree classification after cavitation occurs, lacking the ability to dynamically predict cavitation before it occurs, and failing to meet the urgent needs of real-time intelligent operation and maintenance and risk proactive prevention of equipment in current hydropower stations. SUMMARY

[0005] Therefore, the application aims to provide a support vector machine-based water turbine cavitation prediction method and system to solve the problems in the background art.

[0006] To achieve the above-mentioned purpose, the application adopts the following technical solutions: The support vector machine-based water turbine cavitation prediction method of the application comprises the following steps: Obtain the geometric information of the water turbine and the volute inlet pressure and the draft tube outlet flow rate of the water turbine in real time, wherein the geometric information is obtained from design drawings or three-dimensional full-size scanning results; Perform three-dimensional modeling based on the geometric information of the water turbine to obtain a water turbine three-dimensional simulation model, wherein the three-dimensional simulation model is a fluid mechanics simulation grid model; Construct a turbulent time-averaged motion equation of the water turbine based on the water turbine three-dimensional simulation model, and solve the turbulent time-averaged motion equation to obtain the pressure field and the velocity field of the water turbine three-dimensional simulation model; Extract a plurality of training samples from the pressure field and the velocity field of the water turbine three-dimensional simulation model, and label the training samples to obtain labeled training samples; Train a support vector machine based on the labeled training samples to obtain a cavitation prediction model; Perform cavitation prediction on the water turbine based on the volute inlet pressure, the draft tube outlet flow rate, and the cavitation prediction model to obtain a cavitation prediction result.

[0007] In an embodiment of the application, the mathematical expression of the turbulent time-averaged motion equation is: In the formula, denotes the fluid density, denotes time, denotes the turbulent velocity in the i and j directions, respectively, denotes the turbulent pressure, denotes the spatial position coordinates in the i and j directions, respectively, denotes the gravitational acceleration, denotes the fluid dynamics viscosity, denotes the average turbulent velocity in the i and j directions, respectively, denotes the average turbulent pressure, denotes the turbulent velocity fluctuation value in the i and j directions, respectively.

[0008] In an embodiment of the application, solving the turbulent time-averaged motion equation to obtain the pressure field and the velocity field of the water turbine three-dimensional simulation model comprises: based on the Standard The turbulence model closes the solution of the turbulence time-average motion equation, and selects a target time step and a target calculation total time length in the solution.

[0009] In an embodiment of the present application, a plurality of training samples are extracted from the pressure field and the velocity field of the three-dimensional simulation model of the water turbine; and the training samples are labeled to obtain labeled training samples, including: A plurality of training samples are extracted from the pressure field and the velocity field of the three-dimensional simulation model of the water turbine; and the training samples are labeled to obtain labeled training samples, including: The cavitation index of each training sample is calculated The mathematical expression of the cavitation index is: In the formula, represents the fluid density, represents the minimum pressure of the monitoring area, represents the saturated vapor pressure, represents the maximum flow rate of the monitoring area; Each training sample is labeled based on the value of the cavitation index to obtain a labeled training sample, wherein the label includes no cavitation, incipient cavitation and severe cavitation.

[0010] In an embodiment of the present application, a support vector machine is trained based on the labeled training sample to obtain a cavitation prediction model, including: The pressure and the velocity in the labeled training sample are normalized to obtain an input sample; The input sample is input into a support vector machine model, and the label thereof is trained to obtain a cavitation prediction model.

[0011] In an embodiment of the present application, the mathematical expression of the decision boundary of the cavitation prediction model is: In the formula, represents the normalized pressure value, represents the normalized velocity value, represents the velocity reference value.

[0012] In an embodiment of the present application, the cavitation of the water turbine is predicted based on the volute inlet pressure, the draft tube outlet flow and the cavitation prediction model to obtain a cavitation prediction result, including: The volute inlet pressure is converted into the minimum pressure based on a pre-constructed empirical relationship, and the draft tube outlet flow is converted into the maximum velocity wherein, a mathematical expression of the empirical relationship is: wherein, and are calibration coefficients, represents the inlet pressure of the volute, represents the outlet flow of the draft tube, represents the throat cross-sectional area of the draft tube; the minimum pressure and the maximum velocity are normalized and input into the cavitation prediction model to obtain a cavitation prediction result.

[0013] The application also provides a support vector machine-based water turbine cavitation prediction system, comprising: an acquisition module configured to acquire geometric information of a water turbine and acquire pressure distribution data and velocity distribution data inside the water turbine in real time, wherein the geometric information is obtained from design drawings or three-dimensional full-size scanning results; a modeling module configured to perform three-dimensional modeling based on the geometric information of the water turbine to obtain a three-dimensional simulation model of the water turbine, wherein the three-dimensional simulation model is a fluid mechanics simulation grid model; a solving module configured to construct a turbulent time-averaged motion equation of the water turbine based on the three-dimensional simulation model of the water turbine, and solve the turbulent time-averaged motion equation to obtain a pressure field and a velocity field of the three-dimensional simulation model of the water turbine; a sample extraction module configured to extract a plurality of training samples from the pressure field and the velocity field of the three-dimensional simulation model of the water turbine, and label the training samples to obtain labeled training samples; a training module configured to train a support vector machine based on the labeled training samples to obtain a cavitation prediction model; a prediction module configured to perform cavitation prediction on the water turbine based on the inlet pressure of the volute, the outlet flow of the draft tube and the cavitation prediction model to obtain a cavitation prediction result.

[0014] The support vector machine-based water turbine cavitation prediction method and system of the application directly focus on the internal flow field physical nature of cavitation initiation through the advanced means of numerical simulation, and a mapping model of the corresponding cavitation risk is established through the support vector machine to realize real-time and accurate early warning of the cavitation state. The method uses a full-parameterized modeling and corrected physical model, has the significant advantages of low cost, strong reproducibility and easy popularization in different types and working conditions of water turbines, thereby providing prospective technical support for the safe, stable and efficient operation of the water turbine. BRIEF DESCRIPTION OF DRAWINGS

[0015] The present application will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart illustrating a turbine cavitation prediction method based on support vector machine in one embodiment of this application; Figure 2 This is a structural diagram of a turbine cavitation prediction system based on support vector machine, as shown in one embodiment of this application. Detailed Implementation

[0016] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0017] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the illustrations only show the layers related to this application and are not drawn according to the actual number, shape and size of the layers in the actual implementation. In the actual implementation, the form, number and proportion of each layer can be arbitrarily changed, and the layer layout may also be more complex.

[0018] Numerous details are explored in the following description to provide a more thorough explanation of embodiments of this application; however, it will be apparent to those skilled in the art that embodiments of this application may be practiced without these specific details.

[0019] Figure 1 This is a flowchart illustrating a support vector machine-based method for predicting cavitation in hydroelectric turbines, as shown in one embodiment of this application. Figure 1 As shown, the turbine cavitation prediction method based on support vector machine in this embodiment includes a model training phase and an application phase. The training phase of the support vector machine model is as follows: S110, Obtain the geometric information of the water turbine, wherein the geometric information comes from design drawings or three-dimensional full-size scan results; S120, Based on the geometric information of the water turbine, a three-dimensional model is performed to obtain a three-dimensional simulation model of the water turbine, wherein the three-dimensional simulation model is a fluid dynamics simulation mesh model; First, according to the water turbine design drawings or three-dimensional full-size scanning results, a computational fluid dynamics simulation grid model containing the fluid domain is drawn. The fluid part needs to include the main flow components, including the spiral case, fixed guide vanes, movable guide vanes, runner and draft tube, etc. The grid of the fluid part adopts hexahedral elements and at least 5 layers of boundary layer grid are drawn at the flow boundary. In addition, in order to facilitate the subsequent calculation of fluid added mass, the fluid domain grid of the entire runner part is also divided into tetrahedral elements. During the grid drawing process, the topological structure of the fluid grid at the fluid-structure coupling surface needs to be checked and ensured to be the same, so as to facilitate the subsequent projection of the fluid pressure field.

[0020] S130, based on the water turbine three-dimensional simulation model, a turbulent time-averaged motion equation of the water turbine is constructed, and the turbulent time-averaged motion equation is solved to obtain a pressure field and a velocity field of the water turbine three-dimensional simulation model; Specifically, assuming that the fluid in the water turbine is an incompressible fluid, the constructed turbulent time-averaged motion equation is as follows: In the formula, denotes the fluid density, denotes time, denotes the turbulent velocity in the i and j directions, respectively, denotes the turbulent pressure, denotes the spatial position coordinates in the i and j directions, respectively, denotes the gravitational acceleration, denotes the fluid dynamics viscosity, denotes the average turbulent velocity in the i and j directions, respectively, denotes the average turbulent pressure, denotes the turbulent velocity fluctuation value in the i and j directions, respectively.

[0021] In this application, the finite volume method is used to solve the turbulent time-averaged equation, and specifically, the Standard The turbulent model (a semi-empirical turbulent simulation method widely used in engineering calculation) is used to close the solution of the turbulent time-averaged motion equation. The spiral case inlet flow velocity and the draft tube outlet pressure monitored under the current operating power of the water turbine are used as the boundary conditions of the computational fluid dynamics model, and appropriate time step and total calculation time are selected to solve the steady flow pressure and flow velocity.

[0022] S140, a plurality of training samples are extracted from the pressure field and the velocity field of the water turbine three-dimensional simulation model; and the training samples are labeled to obtain labeled training samples; The training data of the support vector machine model in the application are all from the flow pressure field and the velocity field obtained by simulation (the pressure field includes the pressure values of multiple points in the monitoring area, and the velocity field includes the velocity values of multiple points in the monitoring area). When extracting the training data, the spatial coupling position of the low-pressure area (pressure wherein is the saturated vapor pressure of water) and the high-speed area (velocity wherein is the characteristic reference velocity of the flow field) in the flow field is monitored in real time, so as to extract the training data (the training data include a group of pressure values and velocity values). After extraction, the cavitation index is calculated, and labeling is performed based on the cavitation index. The specific process includes: S141, extracting the pressure and velocity of multiple points in the monitoring area from the pressure field and the velocity field of the three-dimensional simulation model of the water turbine to obtain multiple training samples; wherein the pressure value and the velocity value in each training sample are from multiple detection points in the low-pressure area and the high-speed area of the flow field.

[0023] S142, calculating the cavitation index of each training sample wherein the mathematical expression of the cavitation index is: wherein represents the fluid density, represents the minimum pressure of the monitoring area, represents the saturated vapor pressure, represents the maximum flow velocity of the monitoring area; S143, labeling each training sample based on the value of the cavitation index to obtain labeled training samples, wherein the label includes no cavitation, incipient cavitation and severe cavitation.

[0024] Specifically, when is determined as no cavitation (the label is marked as 0), when is determined as incipient cavitation (the label is marked as 1), and when ≤ 0.15 is determined as severe cavitation (the label is marked as 2). Based on the CFD simulation results, the pressure and velocity data of each monitoring point are extracted, and the cavitation label is corresponded to form a training sample set.

[0025] S150, training the support vector machine based on the labeled training sample to obtain a cavitation prediction model, and the training process includes: S151, normalizing the pressure and velocity in the labeled training sample to obtain an input sample; Specifically, the input pressure and velocity data are standardized to eliminate dimensional differences and improve the stability and convergence of the subsequent classification model.

[0026] The pressure P is normalized by the minimum-maximum normalization formula to a dimensionless value where and are the minimum and maximum values of the pressure data set, respectively. The velocity is normalized by dividing by the characteristic reference velocity of the flow field, resulting in , which brings the data to a comparable scale.

[0027] S152, the input sample is input into the support vector machine model, and its label is collected for training to obtain a cavitation prediction model.

[0028] Specifically, the training of the support vector machine is to classify the normalized pressure-velocity combination data using a supervised learning algorithm to distinguish different cavitation states.

[0029] The classification label is based on the cavitation state verified by experiments, represented by the numerical values 0, 1, and 2 (0 corresponds to no cavitation, 1 corresponds to incipient cavitation, and 2 corresponds to severe cavitation). The kernel function uses a radial basis function (RBF), which has the form where , are two independent feature vectors (sample points) from the data set, is a kernel parameter used to handle non-linear classification problems, and high-precision segmentation is achieved by optimizing the hyperplane. represents the output value of the kernel function, which is a scalar, used to measure , the similarity in the high-dimensional feature space.

[0030] After training, a complex decision boundary is formed inside the SVM model. To facilitate engineering applications, this boundary can be fitted or approximated to obtain the following empirical criterion formula: where represents the normalized pressure value, represents the normalized velocity value, represents the velocity reference value.

[0031] This quadratic polynomial equation describes the critical relationship between the normalized pressure and the normalized velocity, and the coefficients (-0.32, 0.18, 0.05) are determined through model training and can be directly used for cavitation prediction in engineering applications.

[0032] SVM model verification: To evaluate the accuracy and generalization ability of the constructed support vector machine (SVM) model in identifying the cavitation state of a hydro turbine, this step uses a combination of cross-validation and independent test sets for systematic verification.

[0033] Dataset partitioning: The complete sample set generated based on CFD simulation is divided into a training set and an independent test set in an 8:2 ratio. The training set is used for model parameter learning, while the test set is not involved in the training process and is used to objectively evaluate model performance.

[0034] Validation metric definition: The following multi-dimensional indicators are used to quantify classification performance: Accuracy: The proportion of samples that correctly predict the total number of samples. F1-score: The harmonic average of precision and recall, especially suitable for class imbalance scenarios; Confusion matrix: Analyzes the distribution of misclassifications among different categories to identify easily confused operating conditions.

[0035] Verification results: In a mixed-flow turbine case study (3.5 million grid points), 420 sets of valid simulation data were generated. After training, the SVM model performed as follows on the independent test set: The prediction accuracy for "initial cavitation" reached 91.7%; The F1-score of 0.893 indicates that the model achieves a good balance between sensitivity and specificity. Comparative analysis: Compared with traditional methods such as pressure thresholding and vibration signal FFT+BP neural networks, the SVM model proposed in this invention performs better on the same test set: Detecting the initial cavitation trend more than 6 minutes in advance; The false alarm rate was reduced to 8.2%, which is significantly better than the comparison method (average 18.5%). A single inference operation takes less than 20ms, meeting the real-time requirements of engineering projects.

[0036] In summary, the verification results show that the SVM model constructed in this invention has high accuracy, strong robustness and good engineering applicability, and can be reliably used for intelligent identification and early warning of cavitation state in water turbines.

[0037] The cavitation prediction stage for water turbines is as follows: After the validated SVM model parameters are solidified, they are deployed to the online monitoring system. The workflow is as follows: This step deploys the well-verified SVM model to the hydraulic turbine operation monitoring system, building a lightweight, embeddable cavitation real-time prediction module, and realizing the full-process automatic early warning from field data input to risk output.

[0038] The system adopts a four-layer architecture of "data acquisition-feature estimation-model inference-early warning decision", as shown in the following table: Table 1. System architecture description As shown in the above table, in the prediction process in the present application, the input feature value of the prediction is the pressure-velocity combination matrix, and the output is the cavitation degree label (label 0 / 1 / 2), and the specific process includes: S160, converting the spiral casing inlet pressure into the minimum pressure based on the pre-constructed empirical relationship and converting the draft tube outlet flow into the maximum velocity wherein the mathematical expression of the empirical relationship is: wherein, and are calibration coefficients, represents the spiral casing inlet pressure, represents the draft tube outlet flow, represents the draft tube throat cross-sectional area; S170, after normalizing the minimum pressure and the maximum velocity , input them into the cavitation prediction model to obtain the cavitation prediction result.

[0039] The present application obtains the cavitation evolution law under typical working conditions through CFD simulation, and trains a lightweight support vector machine prediction model based on this, realizing the technical transformation from offline analysis to online early warning.

[0040] The innovation of the above process lies in: First, unlike traditional external signal monitoring methods, the cavitation dynamic criterion based on internal pressure, flow rate, vorticity and other parameters is directly constructed, solving the problem of missed judgment and false judgment of traditional single parameter monitoring, accurately identifying the flow field characteristics of cavitation initiation, avoiding false or missed early warning caused by single parameter abnormalities, and improving the timeliness of early warning to some extent; Second, by establishing a numerical simulation model, the technical limitations of traditional methods relying on physical sensors, model tests and hardware systems are broken through, and the universality and economy of the method are improved; Thirdly, the method innovatively combines the support vector machine algorithm with the flow field simulation, realizes the dynamic real-time prediction of the cavitation state by establishing a mapping model between the pressure-velocity combination characteristics and the cavitation risk, and can provide key decision basis for early prevention and control of the cavitation risk.

[0041] The support vector machine-based water turbine cavitation prediction method provided in the application directly focuses on the physical nature of the internal flow field causing cavitation through the advanced means of numerical simulation, establishes a mapping model of the corresponding cavitation risk through a support vector machine, and realizes real-time and accurate early warning of the cavitation state. The method adopts a fully parameterized modeling and corrected physical model, has the significant advantages of low cost, strong reproducibility and easy popularization in different types and working conditions of water turbines, thereby providing prospective technical support for safe, stable and efficient operation of the water turbine.

[0042] As shown in Figure 2 The application further provides a support vector machine-based water turbine cavitation prediction system, which comprises: An acquisition module is configured to acquire geometric information of a water turbine and acquire pressure distribution data and velocity distribution data inside the water turbine in real time, wherein the geometric information is obtained from design drawings or three-dimensional full-size scanning results; A modeling module is configured to perform three-dimensional modeling based on the geometric information of the water turbine to obtain a three-dimensional simulation model of the water turbine, wherein the three-dimensional simulation model is a fluid mechanics simulation grid model; A solving module is configured to construct a turbulent time-averaged motion equation of the water turbine based on the three-dimensional simulation model of the water turbine, and solve the turbulent time-averaged motion equation to obtain a pressure field and a velocity field of the three-dimensional simulation model of the water turbine; A sample extraction module is configured to extract a plurality of training samples from the pressure field and the velocity field of the three-dimensional simulation model of the water turbine, and label the training samples to obtain labeled training samples; A training module is configured to train a support vector machine based on the labeled training samples to obtain a cavitation prediction model; A prediction module is configured to perform cavitation prediction on the water turbine based on the volute inlet pressure, the draft tube outlet flow and the cavitation prediction model to obtain a cavitation prediction result.

[0043] The support vector machine-based water turbine cavitation prediction system provided in the application directly focuses on the physical nature of the internal flow field causing cavitation through the advanced means of numerical simulation, establishes a mapping model of the corresponding cavitation risk through a support vector machine, and realizes real-time and accurate early warning of the cavitation state. The method adopts a fully parameterized modeling and corrected physical model, has the significant advantages of low cost, strong reproducibility and easy popularization in different types and working conditions of water turbines, thereby providing prospective technical support for safe, stable and efficient operation of the water turbine.

[0044] The embodiment also provides an electronic terminal, comprising a processor and a memory. The memory is configured to store a computer program, and the processor is configured to execute the computer program stored in the memory, so that the terminal executes any method in the embodiment.

[0045] Those skilled in the art can understand that all or part of the steps of the methods in the embodiments can be completed by a computer program related hardware. The foregoing computer program can be stored in a computer readable storage medium. When the program is executed, the steps of the foregoing methods are executed; and the foregoing storage medium includes ROM, RAM, magnetic disk or optical disk and various storage media that can store program codes.

[0046] The electronic terminal provided in the embodiment includes a processor, a memory, a transceiver and a communication interface. The memory and the communication interface are connected with the processor and the transceiver and complete communication between each other. The memory is configured to store a computer program, the communication interface is configured to communicate, and the processor and the transceiver are configured to run the computer program, so that the electronic terminal executes each step of the method and deploys the cavitation prediction model as described above.

[0047] In the embodiment, the memory can include random access memory (RAM) and can also include non-volatile memory such as at least one disk memory.

[0048] The processor described above can be a general processor including a central processing unit (CPU), a network processor (NP) and the like; and can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0049] In the foregoing embodiments, although the present application has been described in conjunction with specific embodiments thereof, numerous alternatives, modifications and variations will be readily apparent to those having ordinary skill in the art in the light of the foregoing descriptions. The embodiments of the present application are intended to embrace all such alternatives, modifications and variations as falling within the scope of the appended claims.

[0050] The foregoing embodiments are merely illustrative of the principles of the application and the efficacy thereof, and are not intended to limit the application. Any modification or alteration of the foregoing embodiments that will not depart from the spirit and scope of the application is intended to be encompassed by the claims of the application.

Claims

1. A method for predicting cavitation in hydro-turbines based on support vector machines, characterized in that, Including the following steps: The geometric information of the turbine is obtained, and the turbine's volute inlet pressure and tailrace outlet flow rate are obtained in real time. The geometric information comes from design drawings or three-dimensional full-size scanning results. Based on the geometric information of the water turbine, a three-dimensional model is created to obtain a three-dimensional simulation model of the water turbine, wherein the three-dimensional simulation model is a fluid dynamics simulation mesh model; Based on the three-dimensional simulation model of the turbine, the turbulent time-averaged motion equation of the turbine is constructed, and the turbulent time-averaged motion equation is solved to obtain the pressure field and velocity field of the three-dimensional simulation model of the turbine. Multiple training samples are extracted from the pressure field and velocity field of the three-dimensional simulation model of the water turbine; and the training samples are labeled to obtain labeled training samples. The support vector machine is trained based on the labeled training samples to obtain a cavitation prediction model. Based on the inlet pressure of the vortex casing, the outlet flow rate of the tailrace pipe, and the cavitation prediction model, cavitation prediction of the turbine is performed to obtain cavitation prediction results.

2. The turbine cavitation prediction method based on support vector machine according to claim 1, characterized in that, The mathematical expression for the time-averaged motion equation of the turbulence is: In the formula, Indicates fluid density, Indicates time, These represent the turbulent velocities in the i and j directions, respectively. Indicates turbulent pressure. These represent the spatial coordinates in the i and j directions, respectively. Represents gravitational acceleration. Indicates fluid dynamic viscosity, Let i and j represent the average turbulent velocities, respectively. Indicates the average turbulent pressure. These represent the turbulent velocity fluctuations in the i-direction and the j-direction, respectively.

3. The turbine cavitation prediction method based on support vector machine according to claim 1, characterized in that, Solving the time-averaged turbulence equations yields the pressure and velocity fields of the three-dimensional simulation model of the turbine, including: Based on Standard The turbulence model solves the time-averaged motion equations of the turbulence in a closed loop, and selects the pressure field and velocity field with the target time step and the target total calculation time in the solution.

4. The turbine cavitation prediction method based on support vector machine according to claim 1, characterized in that, Multiple training samples were extracted from the pressure field and velocity field of the three-dimensional simulation model of the water turbine. The training samples are then labeled to obtain labeled training samples, including: Pressure and velocity data at multiple points within the monitoring area are extracted from the pressure and velocity fields of the three-dimensional simulation model of the water turbine to obtain multiple training samples. Calculate the cavitation index for each training sample The mathematical expression for the cavitation index is: In the formula, Indicates fluid density, Indicates the minimum pressure in the monitored area. Indicates saturated vapor pressure. Indicates the maximum flow velocity in the monitored area; Based on the cavitation index The value is used to label each training sample to obtain labeled training samples, wherein the labels include no vacuolation, nascent vacuolation, and severe vacuolation.

5. The turbine cavitation prediction method based on support vector machine according to claim 1, characterized in that, The support vector machine is trained based on the labeled training samples to obtain a cavitation prediction model, including: The pressure and velocity in the labeled training samples are normalized to obtain the input samples; The input samples are fed into a support vector machine model and their labels are used for training to obtain a cavitation prediction model.

6. The turbine cavitation prediction method based on support vector machine according to claim 5, characterized in that, The mathematical expression for the decision boundary of the cavitation prediction model is: In the formula, This represents the normalized pressure value. This represents the normalized velocity value. This indicates a speed reference value.

7. The turbine cavitation prediction method based on support vector machine according to claim 1, characterized in that, Based on the inlet pressure of the vortex casing, the outlet flow rate of the draft tube, and the cavitation prediction model, cavitation prediction is performed on the turbine to obtain cavitation prediction results, including: The inlet pressure of the vortex casing is converted into a minimum pressure based on a pre-established empirical formula. And convert the tailwater outlet flow rate into the maximum velocity. The mathematical expression of the empirical relation is as follows: In the formula, and All are calibration coefficients. This indicates the inlet pressure of the volute. Indicates the flow rate at the tailrace pipe outlet. This indicates the cross-sectional area of ​​the tailpipe throat; For the minimum pressure and the maximum speed After normalization, the result is input into the cavitation prediction model to obtain the cavitation prediction result.

8. A turbine cavitation prediction system based on support vector machine, characterized in that, include: The acquisition module is used to acquire the geometric information of the water turbine and acquire the pressure distribution data and velocity distribution data inside the water turbine in real time. The geometric information comes from the design drawings or the results of three-dimensional full-size scanning. The modeling module is used to perform three-dimensional modeling based on the geometric information of the water turbine to obtain a three-dimensional simulation model of the water turbine, wherein the three-dimensional simulation model is a fluid dynamics simulation mesh model; The solution module is used to construct the turbulent time-averaged motion equation of the turbine based on the three-dimensional simulation model of the turbine, and solve the turbulent time-averaged motion equation to obtain the pressure field and velocity field of the three-dimensional simulation model of the turbine. The sample extraction module is used to extract multiple training samples from the pressure field and velocity field of the three-dimensional simulation model of the water turbine; and to label the training samples to obtain labeled training samples. The training module is used to train the support vector machine based on the labeled training samples to obtain the cavitation prediction model. The prediction module is used to predict the cavitation of the turbine based on the inlet pressure of the vortex casing, the outlet flow rate of the tailrace pipe, and the cavitation prediction model, and obtain the cavitation prediction result.

9. An electronic device, characterized in that, include: Processor and memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to deploy the cavitation prediction model as described in claims 1-7, so that the electronic device performs the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images