Centrifugal pump performance degradation evaluation method based on twin data

Abstract

The invention discloses a centrifugal pump performance degradation evaluation method based on twin data. The method comprises the steps that operation data of a centrifugal pump are collected to obtain actually-measured vibration signals; constructing a centrifugal pump digital twinning model in a service state, and generating twinning vibration signals of the centrifugal pump in different health states; decomposing twin vibration signals of the centrifugal pump in different health states to obtain a product function component; determining a candidate product function component according to the similarity between the product function component and the original vibration signal; extracting a candidate singular value component corresponding to each candidate product function component; taking each candidate singular value component as the input of a rapid search clustering model, and determining a clustering center; and introducing a quantitative index confidence function to determine a degradation index, and judging the degradation state of the centrifugal pump according to the degradation index. By introducing the digital twin technology and the advanced algorithm, the automation degree and accuracy of evaluation are improved, and a novel and effective solution is provided for health management of the centrifugal pump.

Description

Technical Field

[0001] The present invention relates to the field of intelligent industrial technology, and in particular to a centrifugal pump performance degradation assessment method based on twin data. Background Art

[0002] Centrifugal pumps are common, critical equipment in the machinery industry, widely used in complex systems such as large-scale nuclear power plants, water supply and drainage, and marine propulsion. Their efficient and reliable operation is crucial for system safety and stability. Therefore, research on centrifugal pump performance degradation assessment methods is of great significance for ensuring the safe and efficient operation of complex systems and has broad engineering application prospects.

[0003] At present, traditional performance degradation assessment modeling methods, such as single or multiple time-frequency domain indicators (such as root mean square value, kurtosis and kurtosis, etc.) and signal decomposition models (wavelet transform, empirical mode decomposition, etc.), all require rich manual experience (such as selecting appropriate wavelet functions and fault frequency bands for decomposition) to select appropriate feature indicators and cross-integrate with models, and the above modeling methods are relatively cumbersome.

[0004] In response to the problems existing in traditional performance degradation assessment methods, the present invention proposes a centrifugal pump performance degradation assessment method based on twin data, which reduces the dependence on manual experience, and at the same time solves the problem that the signal performance degradation state is difficult to divide due to continuous changes, thereby saving economic costs. Summary of the Invention

[0005] In view of this, the present invention provides a centrifugal pump performance degradation assessment method based on twin data to solve the technical problems in the existing centrifugal pump performance degradation assessment technology, which requires rich manual experience to select appropriate characteristic indicators to judge the degradation situation, resulting in low assessment efficiency and too strong dependence on subjectivity.

[0006] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a method for evaluating centrifugal pump performance degradation based on twin data, comprising:

[0008] Collect the operating data of the centrifugal pump in real time and obtain the measured vibration signal based on the operating data;

[0009] Based on the centrifugal pump's physical entity information, operating conditions, and measured vibration signals, a digital twin model of the centrifugal pump in service is constructed, generating twin vibration signals of the centrifugal pump in different health states under service conditions.

[0010] The twin vibration signals of the centrifugal pump under different health conditions are decomposed using the local mean decomposition method to obtain product function components. The similarity between each product function component and the original vibration signal is calculated. The product function components are sorted based on the similarity and several candidate product function components with the highest similarity are selected.

[0011] The singular value decomposition method is used to extract the candidate singular value components corresponding to each candidate product function component;

[0012] Each candidate singular value component is used as input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states;

[0013] The eigenvector is constructed according to the candidate singular value components, and the quantitative index confidence function is introduced. The distance between the eigenvector and the cluster center is used as the degradation index, and the degradation state of the centrifugal pump is judged according to the degradation index.

[0014] Furthermore, the distance between the feature vector and the cluster center is used as a degradation indicator, which is expressed as follows:

[0015]

[0016] Where CV represents the similarity between the i-th candidate singular value and the cluster center under normal circumstances; D represents the distance between any two candidate singular values; and S represents the scaling factor related to the cluster center.

[0017] Furthermore, based on the centrifugal pump's physical entity information, operating conditions, and measured vibration signals, a digital twin model of the centrifugal pump in service is constructed, including:

[0018] Based on the physical entity information and operating conditions of the centrifugal pump, a finite element model of the centrifugal pump is established using finite element analysis software, and simulation data is generated;

[0019] Use Kalman filtering to update and correct the simulation data of the centrifugal pump finite element model through measured vibration signals to build a digital twin model of the centrifugal pump;

[0020] The digital twin model of the centrifugal pump is used to generate twin vibration signals of the centrifugal pump in different health states under service conditions.

[0021] Furthermore, the finite element model of the centrifugal pump is established using finite element analysis software, including:

[0022] Import the centrifugal pump geometry model into the finite element analysis software and define the material properties for each part of the geometry model;

[0023] The geometric model is divided into multiple basic units using finite element analysis software, and the properties of the mesh are adjusted to ensure simulation accuracy and computational efficiency;

[0024] Apply boundary conditions and loads to the geometric model based on the operating conditions of the centrifugal pump to simulate the actual operating state;

[0025] The model response information is solved according to the simulation model, and simulation data corresponding to the measured data is generated.

[0026] Furthermore, each candidate singular value component is used as input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states, including:

[0027] A sample data set is formed based on each candidate singular value component at different running times, and the Euclidean distance between any two samples in the data set is calculated;

[0028] Calculate the local density of each element based on the Euclidean distance, sort the local density of each data in the sequence from large to small, and calculate the judgment distance of each data point;

[0029] Assign a cluster value to each data point based on local density and decision distance;

[0030] The cluster centers are determined based on the arranged cluster values. The elements with the highest cluster values ​​are prioritized as potential cluster centers. The change points where the cluster values ​​suddenly change are identified and used as cluster centers of different degradation states to determine the number of clusters and the location of the center points.

[0031] The data points are automatically divided into clusters of different degradation states according to whether the distance between the data points and the cluster centers of different degradation states exceeds the preset cutoff distance.

[0032] Furthermore, the local density of each element is calculated based on the Euclidean distance, which is expressed as follows:

[0033]

[0034] in, is the Euclidean distance between data points i and j; The preset cutoff distance.

[0035] Furthermore, the local density of each data point in the sequence is sorted in descending order, and the determination distance of each data point is calculated, including:

[0036] If the local density of a data point is the highest, the judgment distance of the data point is the distance from the data point to the farthest point in the data set; for other points where the local density is not the highest, the judgment distance is the distance to the nearest point with a higher density than the point; it can be expressed as:

[0037]

[0038] Among them, the data points are arranged in descending order according to the local density. When i=1, is the data point with the highest local density; Representation and data points With data points The Euclidean distance between Represents data points Compared with High density of data points The Euclidean distance between .

[0039] In a second aspect, the present invention further provides a centrifugal pump performance degradation assessment system based on twin data, comprising:

[0040] The data acquisition module is used to collect the operating data of the centrifugal pump in real time and obtain the measured vibration signal based on the operating data;

[0041] The twin model construction module is used to build a digital twin model of the centrifugal pump in service based on the centrifugal pump's physical entity information, operating conditions, and measured vibration signals, and to generate twin vibration signals of the centrifugal pump in different health states under service conditions.

[0042] The signal decomposition module is used to decompose the twin vibration signals of the centrifugal pump under different health conditions using the local mean decomposition method to obtain product function components; calculate the similarity between each product function component and the original vibration signal; sort the product function components based on the similarity and select several candidate product function components with the highest similarity;

[0043] A feature extraction module is used to extract candidate singular value components corresponding to each candidate product function component by using a singular value decomposition method;

[0044] A clustering module is used to use each candidate singular value component as input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states;

[0045] The evaluation module is used to construct a eigenvector based on the singular value components, introduce a quantitative indicator confidence function, use the distance between the eigenvector and the cluster center as a degradation indicator, and judge the degradation state of the centrifugal pump based on the degradation indicator.

[0046] In a third aspect, the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the centrifugal pump performance degradation assessment method based on twin data as described in any one of the above technical solutions.

[0047] Compared with the existing technology, the centrifugal pump performance degradation assessment method based on twin data proposed in this invention has the following advantages:

[0048] (1) Taking into full consideration the fact that centrifugal pumps are mostly in normal state in actual engineering applications and lack abnormal sample data, digital twin technology is introduced to generate twin vibration signals to simulate the performance of centrifugal pumps in different health states, effectively expanding the number of samples and ensuring a comprehensive performance evaluation under normal conditions.

[0049] (2) The product function (PF) components extracted by local modal decomposition (LMD) combined with the singular values ​​(SVs) extracted by singular value decomposition (SVD) in step 4 provide rich feature information for performance degradation assessment. This comprehensive feature extraction method can more comprehensively reflect the health status of the centrifugal pump and improve the accuracy of the assessment.

[0050] (3) The fast clustering search (CFS) algorithm is used to automatically divide sample labels, which can accurately identify and classify different degradation states (normal state, initial degradation, and severe degradation), reducing the dependence on manual experience and improving the objectivity and accuracy of the evaluation.

[0051] The centrifugal pump performance degradation assessment method based on twin data of the present invention not only solves the problem of insufficient samples, but also improves the degree of automation and accuracy of the assessment by introducing advanced algorithms and technologies, providing a novel and effective solution for the health management of centrifugal pumps. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic flow chart of a centrifugal pump performance degradation assessment method based on twin data provided by the present invention;

[0053] Figure 2 A schematic diagram of the application of the method provided by the present invention;

[0054] Figure 3 This is a schematic diagram of the structure of the centrifugal pump performance degradation assessment system based on twin data provided by the present invention. DETAILED DESCRIPTION

[0055] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, and are not used to limit the scope of the present invention.

[0056] See Figure 1 This embodiment provides a centrifugal pump performance degradation assessment method based on twin data, including:

[0057] Step S101: collecting the operating data of the centrifugal pump in real time, and obtaining a measured vibration signal based on the operating data;

[0058] Step S102: constructing a digital twin model of the centrifugal pump in service based on the physical entity information, operating conditions, and measured vibration signals of the centrifugal pump, and generating twin vibration signals of the centrifugal pump in different health states under service conditions;

[0059] Step S103: Decomposing the twin vibration signals of the centrifugal pump under different health states by the local mean decomposition method to obtain product function components; calculating the similarity between each product function component and the original vibration signal; sorting the product function components based on the similarity, and selecting several candidate product function components with the highest similarity;

[0060] Step S104: using the singular value decomposition method to extract the candidate singular value components corresponding to each candidate product function component;

[0061] Step S105: using each candidate singular value component as input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states;

[0062] Step S106: constructing a feature vector based on the candidate singular value components, introducing a quantitative index confidence function, taking the distance between the feature vector and the cluster center as a degradation index, and judging the degradation state of the centrifugal pump based on the degradation index.

[0063] The method of this embodiment collects the operating data of the centrifugal pump in real time to obtain the operating status of the equipment in a timely manner, ensuring real-time monitoring of the equipment performance; by constructing a digital twin model of the centrifugal pump, it can simulate the behavior of the real equipment and generate twin vibration signals under different health states, which helps to fully understand the performance of the equipment under various working conditions; by calculating the similarity between the product function components and the original signal, it can quickly screen out the features closest to the actual state, improve the effectiveness of the evaluation, use the CFS clustering algorithm to achieve automatic division of sample labels, and reduce dependence on manual experience; introduce a quantitative indicator confidence function, and use the distance between the feature vector and the cluster center as a degradation indicator, making the degradation assessment more objective and quantifiable. The method of this embodiment can achieve early warning of equipment, reduce the risk of failure, and reduce maintenance costs.

[0064] As a specific embodiment, in step S101, the method for collecting centrifugal pump operating data is as follows: based on the original plant SCADA system, by installing vibration sensors on both sides of the centrifugal pump unit, using OPC UA technology to collect the vertical and horizontal vibration signals of the equipment and extract the pressure signal from the SCADA system; collect once every 1 second and store it as a sample, and each sample records 10 measured vibration data.

[0065] As a preferred embodiment, in step S102, a digital twin model of the centrifugal pump in service is constructed based on the physical entity information, operating conditions, and measured vibration signals of the centrifugal pump, including:

[0066] Based on the physical entity information and operating conditions of the centrifugal pump, a finite element model of the centrifugal pump is established using finite element analysis software, and simulation data is generated;

[0067] Use Kalman filtering to update and correct the simulation data of the centrifugal pump finite element model through measured vibration signals to build a digital twin model of the centrifugal pump;

[0068] The digital twin model of the centrifugal pump is used to generate twin vibration signals of the centrifugal pump in different health states under service conditions.

[0069] As a preferred embodiment, a finite element model of a centrifugal pump is established using finite element analysis software, including:

[0070] Import the centrifugal pump geometry model into the finite element analysis software and define the material properties for each part of the geometry model;

[0071] The geometric model is divided into multiple basic units using finite element analysis software, and the properties of the mesh are adjusted to ensure simulation accuracy and computational efficiency;

[0072] Apply boundary conditions and loads to the geometric model based on the operating conditions of the centrifugal pump to simulate the actual operating state;

[0073] The model response information is solved according to the simulation model, and simulation data corresponding to the measured data is generated.

[0074] As a specific example, the detailed steps for building a digital twin model are as follows:

[0075] Step 1: Create a geometric model: Import the 3D model created in Solidworks into COMSOL (finite element software) and define material properties for each part in the geometric model, such as elastic modulus, Poisson's ratio, density, thermal conductivity, etc.

[0076] Step 2: Meshing: The geometric model is automatically divided into a series of small units (such as triangles, quadrilaterals, tetrahedrons or hexahedrons) using COMSOL software, and the size, shape and distribution of the mesh are manually adjusted to ensure accuracy and computational efficiency in key areas.

[0077] Step 3: Boundary conditions and loads: Based on the actual working conditions (the operating data collected in step S101), apply appropriate boundary conditions and loads to the model to ensure that they conform to the actual conditions and simulate the state of the analysis object as accurately as possible.

[0078] Step 4: Generate simulation data: Based on the defined model, material properties, mesh division, boundary conditions and load information, solve the model response (such as displacement, stress, temperature, etc.) and generate simulation data similar to the measured data.

[0079] Step 5: Build a digital twin model: Use the Kalman filter method to update and correct the simulation data of the centrifugal pump finite element model using the measured data of normal operation to build a digital twin model of the centrifugal pump;

[0080] Finally, the constructed digital twin model of the centrifugal pump is used to generate twin vibration signals of the centrifugal pump in different health states under service conditions.

[0081] Through the above steps, the constructed digital twin model of a centrifugal pump can generate twin vibration signals for the centrifugal pump under different service conditions and health states. Even in actual engineering applications, centrifugal pumps are mostly in normal conditions and lack abnormal sample data. However, the digital twin model can simulate vibration signals under various possible abnormal health states based on physical principles and model predictions, thus supplementing the lack of abnormal sample data. This provides complete and comprehensive data support for subsequent vibration signal analysis, fault diagnosis, and health status assessment, avoiding the problem of inaccurate or incomplete analysis caused by a lack of abnormal sample data.

[0082] Based on complete data including normal and abnormal conditions, the abnormal vibration signals and changes in the health status of the centrifugal pump can be identified more accurately, which improves the accuracy and reliability of fault diagnosis and is of great significance for timely detection and treatment of potential faults of the centrifugal pump.

[0083] As a specific example, in step S103, LMD decomposition is performed on the twin vibration signals of the centrifugal pump in different health states. In actual calculations, the first two product function (PF) components are generally selected. Specifically:

[0084] First, the LMD method is used to extract the product function (PF) components of the twin vibration signals of the centrifugal pump at different health states. Then, the similarity between each PF component and the original vibration signal is calculated using the correlation coefficient. Finally, the first two PF components are selected based on the correlation coefficient value.

[0085] It should be noted that LMD decomposition (Local Mean Decomposition) is a signal processing technique used to decompose complex signals into several simple components. Its core idea is to decompose the signal into several Intrinsic Mode Functions (IMFs) and a residual using the local mean method. Each IMF represents a specific frequency component of the signal. The PF is usually a combination of IMFs, reflecting the interaction between different frequency components in the signal.

[0086] The correlation function uses the Pearson correlation coefficient, which measures the linear relationship between two variables and has a value range of -1 to 1. Based on the calculated correlation coefficient, the first two PF components with the highest similarity to the original vibration signal are selected. These two PF components have the highest correlation coefficient values. For example, PF1 might be IMF1 + 0.5* IMF2, and PF2 might be 0.8 * IMF3 - IMF4.

[0087] As a specific embodiment, in step S104, the singular value decomposition (SVD) algorithm is used to extract the singular values ​​(SV) corresponding to the first two PF components in step 3, where PF1 corresponds to SV1 and PF2 corresponds to SV2. Singular value decomposition is a matrix decomposition method that can decompose an arbitrary matrix A into the product of three matrices, which can be expressed as follows:

[0088]

[0089] in, is a column-orthogonal matrix containing left singular vectors, Σ is a diagonal matrix containing singular values ​​(SVs), which are non-negative and arranged in descending order; is the transposed matrix of the right singular vectors.

[0090] By performing SVD on the first two PF components, we obtain two singular values, SV1 and SV2. These two singular values ​​can form a vector or matrix. Assuming we have multiple samples (e.g., PF components at different time points), we can organize the singular values ​​of each sample into a vector matrix X. The first two SV components are used as input to the Clustering by Fast Search (CFS) algorithm, which automatically determines the number of clusters and their centers for different degradation states (normal, initial degradation, and severe degradation). Specifically, assuming there are N samples, then:

[0091] (1) Calculation data set The Euclidean distance between any two samples in is the vector matrix of all SV components obtained in step S104.

[0092] (2) Calculate data points The local density ,

[0093]

[0094] in, is the Euclidean distance between data i and j; is the cutoff distance, which is usually given manually. This method uses genetic algorithm to optimize the setting The value is 1.5% (that is, the number of adjacent points accounts for 1.5% of the total number of data points).

[0095] (3) Local density of each data in the sequence Sort in descending order; then calculate the judgment distance , determine the distance is calculated as follows:

[0096] If the local density of a data point is the highest, the judgment distance of the data point is the distance from the data point to the farthest point in the data set; for other points where the local density is not the highest, the judgment distance is the distance to the nearest point with a higher density than the point; it can be expressed as:

[0097]

[0098] Among them, the data points are arranged in descending order according to the local density. When i=1, is the data point with the highest local density; Representation and data points With data points The Euclidean distance between Represents a data point Compared with High density of data points The Euclidean distance between .

[0099] (4) Following the calculation process of steps (2) and (3), assign a value, and then sort in descending order.

[0100] (5) According to each data point The values ​​determine the cluster centers. The values ​​are sorted in descending order, and points with higher values ​​are prioritized as potential cluster centers. Data points with sudden changes in value are identified and selected as cluster centers of different states.

[0101] (6) Whether the distance between the data point and the cluster center of different degradation states exceeds the cutoff distance , which automatically divides the data points into clusters of different degradation states.

[0102] Finally, a quantitative indicator confidence value (CV) function is introduced to evaluate the feature vectors [SV1, SV2] obtained in real time in the step, and the distance between the feature vector and the cluster center point determined by the clustering model S105 is used as the performance degradation indicator.

[0103] Please refer to Figure 2 , the flow chart of the method in practical application is as follows Figure 2 shown.

[0104] like Figure 3 As shown, an embodiment of the present invention further provides a centrifugal pump performance degradation assessment system 300 based on twin data, comprising:

[0105] The data acquisition module 301 is used to collect the operating data of the centrifugal pump in real time and obtain the measured vibration signal based on the operating data;

[0106] The twin model construction module 302 is used to construct a digital twin model of the centrifugal pump in service based on the physical entity information, operating conditions, and measured vibration signals of the centrifugal pump, and generate twin vibration signals of the centrifugal pump in different health states under service conditions;

[0107] The signal decomposition module 303 is configured to decompose the twin vibration signals of the centrifugal pump under different health states using the local mean decomposition method to obtain product function components; calculate the similarity between each product function component and the original vibration signal; sort the product function components based on the similarity, and select several candidate product function components with the highest similarity;

[0108] A feature extraction module 304 is configured to extract candidate singular value components corresponding to each candidate product function component using a singular value decomposition method;

[0109] A clustering module 305 is configured to use each candidate singular value component as an input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states;

[0110] The evaluation module 306 is used to construct a feature vector based on the singular value components, introduce a quantitative index confidence function, use the distance between the feature vector and the cluster center as a degradation index, and judge the degradation state of the centrifugal pump based on the degradation index.

[0111] As a preferred embodiment, the data acquisition module 201 includes a plant monitoring module, a sensor module and a data transmission module;

[0112] The plant monitoring module is used to obtain the pressure signal of the centrifugal pump unit when it is running;

[0113] The sensor module is a vibration displacement sensor installed on the rolling bearing seats on both sides of the centrifugal pump unit, which is used to obtain the vibration signals of the centrifugal pump in the vertical and horizontal directions;

[0114] The data transmission module is used to obtain the operating data of the centrifugal pump using the OPC UA communication method.

[0115] It should be noted that OPC UA's cross-platform nature (supporting Windows / Linux / embedded systems) adapts to the diverse hardware environments of factories and stations. In addition, the OPC UA client subscribes to data point changes, and vibration data is updated and pushed to SCADA in milliseconds, avoiding polling delays.

[0116] This embodiment also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the centrifugal pump performance degradation assessment method based on twin data as described in any of the above technical solutions.

[0117] The computer-readable storage medium and computing device provided according to the above embodiments of the present invention can be implemented with reference to the specific description of the centrifugal pump performance degradation assessment method based on twin data as described above according to the present invention, and have similar beneficial effects as the centrifugal pump performance degradation assessment method based on twin data as described above, which will not be repeated here.

[0118] The twin data-based centrifugal pump performance degradation assessment method disclosed in the present invention can obtain the operating status of the equipment in a timely manner by collecting the operating data of the centrifugal pump in real time, ensuring real-time monitoring of the equipment performance; by constructing a digital twin model of the centrifugal pump, it can simulate the behavior of the real equipment and generate twin vibration signals under different health states, which helps to fully understand the performance of the equipment under various working conditions; by calculating the similarity between the product function components and the original signal, it can quickly screen out the features closest to the actual state, improving the effectiveness of the assessment, and using the CFS clustering algorithm to achieve automatic division of sample labels, reducing dependence on manual experience; introducing a quantitative indicator confidence function, and using the distance between the feature vector and the cluster center as a degradation indicator, making the degradation assessment more objective and quantifiable. The present invention achieves early warning of equipment, reduces the risk of failure, and reduces maintenance costs.

[0119] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A centrifugal pump performance degradation assessment method based on twin data, characterized in that: include: Collect the operating data of the centrifugal pump in real time and obtain the measured vibration signal based on the operating data; Based on the centrifugal pump's physical entity information, operating conditions, and measured vibration signals, a digital twin model of the centrifugal pump in service is constructed, generating twin vibration signals of the centrifugal pump in different health states under service conditions. The twin vibration signals of centrifugal pumps in different health states are decomposed by local mean decomposition method to obtain product function components. Calculate the similarity between each product function component and the original vibration signal; Sort the product function components based on similarity, and select several candidate product function components with the highest similarity; The singular value decomposition method is used to extract the candidate singular value components corresponding to each candidate product function component; Each candidate singular value component is used as input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states; The eigenvector is constructed according to the candidate singular value components, and the quantitative index confidence function is introduced. The distance between the eigenvector and the cluster center is used as the degradation index, and the degradation state of the centrifugal pump is judged according to the degradation index.

2. The centrifugal pump performance degradation assessment method based on twin data according to claim 1, characterized in that: The distance between the eigenvector and the cluster center is used as the degradation index, which is expressed as follows: Where CV represents the similarity between the i-th candidate singular value and the cluster center under normal circumstances; D represents the distance between any two candidate singular values; and S represents the scaling factor related to the cluster center.

3. The centrifugal pump performance degradation assessment method based on twin data according to claim 1, characterized in that: Based on the centrifugal pump's physical entity information, operating conditions, and measured vibration signals, a digital twin model of the centrifugal pump in service is constructed, including: Based on the physical entity information and operating conditions of the centrifugal pump, a finite element model of the centrifugal pump is established using finite element analysis software, and simulation data is generated; Use Kalman filtering to update and correct the simulation data of the centrifugal pump finite element model through measured vibration signals to build a digital twin model of the centrifugal pump; The digital twin model of the centrifugal pump is used to generate twin vibration signals of the centrifugal pump in different health states under service conditions.

4. The centrifugal pump performance degradation assessment method based on twin data according to claim 3 is characterized in that: The finite element model of the centrifugal pump is established using finite element analysis software, including: Import the centrifugal pump geometry model into the finite element analysis software and define the material properties for each part of the geometry model; The geometric model is divided into multiple basic units using finite element analysis software, and the properties of the mesh are adjusted to ensure simulation accuracy and computational efficiency; Apply boundary conditions and loads to the geometric model based on the operating conditions of the centrifugal pump to simulate the actual operating state; The model response information is solved according to the simulation model, and simulation data corresponding to the measured data is generated.

5. The centrifugal pump performance degradation assessment method based on twin data according to claim 1, characterized in that: Each candidate singular value component is used as input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states, including: A sample data set is formed based on each candidate singular value component at different running times, and the Euclidean distance between any two samples in the data set is calculated; Calculate the local density of each element based on the Euclidean distance, sort the local density of each data in the sequence from large to small, and calculate the judgment distance of each data point; Assign a cluster value to each data point based on local density and decision distance; The cluster centers are determined based on the arranged cluster values. The elements with the highest cluster values ​​are prioritized as potential cluster centers. The change points where the cluster values ​​suddenly change are identified and used as cluster centers of different degradation states to determine the number of clusters and the location of the center points. The data points are automatically divided into clusters of different degradation states according to whether the distance between the data points and the cluster centers of different degradation states exceeds the preset cutoff distance.

6. The centrifugal pump performance degradation assessment method based on twin data according to claim 5, characterized in that: The local density of each element is calculated based on the Euclidean distance, which is expressed as follows: in, is the Euclidean distance between data points i and j; The preset cutoff distance.

7. The centrifugal pump performance degradation assessment method based on twin data according to claim 5, characterized in that: The local density of each data point in the sequence is sorted in descending order, and the determination distance of each data point is calculated, including: If the local density of a data point is the highest, the judgment distance of the data point is the distance from the data point to the farthest point in the data set; for other points where the local density is not the highest, the judgment distance is the distance to the nearest point with a higher density than the point; it can be expressed as: Among them, the data points are arranged in descending order according to the local density. When i=1, is the data point with the highest local density; Representation and data points With data points The Euclidean distance between Represents a data point Compared with High density of data points The Euclidean distance between .

8. A centrifugal pump performance degradation assessment system based on twin data, characterized in that: include: The data acquisition module is used to collect the operating data of the centrifugal pump in real time and obtain the measured vibration signal based on the operating data; The twin model construction module is used to build a digital twin model of the centrifugal pump in service based on the centrifugal pump's physical entity information, operating conditions, and measured vibration signals, and to generate twin vibration signals of the centrifugal pump in different health states under service conditions. The signal decomposition module is used to decompose the twin vibration signals of the centrifugal pump under different health states by using the local mean decomposition method to obtain product function components; Calculate the similarity between each product function component and the original vibration signal; Sort the product function components based on similarity, and select several candidate product function components with the highest similarity; A feature extraction module is used to extract candidate singular value components corresponding to each candidate product function component by using a singular value decomposition method; A clustering module is used to use each candidate singular value component as input to a fast search clustering model to determine the number and cluster centers of clusters under different degradation states; The evaluation module is used to construct a eigenvector based on the singular value components, introduce a quantitative indicator confidence function, use the distance between the eigenvector and the cluster center as a degradation indicator, and judge the degradation state of the centrifugal pump based on the degradation indicator.

9. The centrifugal pump performance degradation assessment system based on twin data according to claim 8, characterized in that: The data acquisition module includes a plant monitoring module, a sensor module and a data transmission module; The plant monitoring module is used to obtain the pressure signal of the centrifugal pump unit when it is running; The sensor module is a vibration displacement sensor installed on the rolling bearing seats on both sides of the centrifugal pump unit, which is used to obtain the vibration signals of the centrifugal pump in the vertical and horizontal directions; The data transmission module is used to obtain the operating data of the centrifugal pump using the OPC UA communication method.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the centrifugal pump performance degradation assessment method based on twin data as described in any one of claims 1 to 7 is implemented.

Images