A big data-based pumped storage power station unit state monitoring system and method

Abstract

The application discloses a kind of based on big data pumped storage power station unit condition monitoring system and method, by setting multiple physical property monitoring points and multiple vibration signal monitoring points in pumped storage power station unit monitoring area, and the physical property data and vibration signal data of pumped storage power station unit are collected in real time, while the physical property data and vibration signal data collected are preprocessed by data preprocessing method, after preprocessing, the physical property data and vibration signal data after preprocessing are analyzed by data analysis method, while based on the physical property data and vibration signal data after analysis, fault diagnosis model is constructed and fault grade is determined, finally, based on the fault diagnosis model constructed, the fault real-time prediction model is constructed by digital twin method, and based on the fault real-time prediction model constructed, the state of pumped storage power station unit is monitored and early warning in real time, improve the accuracy of water storage power station unit condition monitoring.

Description

Technical Field

[0001] The present invention relates to the field of monitoring and control technology, and in particular to a system and method for monitoring the status of a pumped storage power station unit based on big data. Background Art

[0002] With the continuous growth of electricity demand and the increasing complexity of grid loads, hydropower units are required to participate in peak load and frequency regulation tasks more frequently. This places higher demands on the performance and stability of hydropower units and also increases the difficulty of unit maintenance and overhaul.

[0003] The prior art, such as the invention patent application with announcement number: CN115437301A, discloses a water treatment unit working status monitoring and control system, which includes: a historical monitoring module, a real-time monitoring module, an intelligent control module and an environmental analysis module. The historical monitoring module is used to analyze the historical usage of the water treatment unit, the real-time monitoring module is used to monitor the operation of the water treatment unit in real time, obtain the operation deviation value of the water treatment unit or generate a normal operation signal, the environmental analysis module is used to analyze the operating environment of the water treatment unit, generate a normal environment signal or obtain the environmental deviation value of the water treatment unit, and the intelligent control module is used to perform intelligent control of the water treatment unit and generate a normal monitoring signal or an abnormal monitoring signal.

[0004] From the above solutions, it can be seen that the current monitoring of hydropower units often focuses on the performance of hydropower units, ignores the stability of hydropower units, and does not adequately monitor the status of hydropower units. Summary of the Invention

[0005] The purpose of the present invention is to provide a pumped storage power station unit status monitoring system and method based on big data, which solves the problems existing in the background technology.

[0006] To solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a method for monitoring the status of a pumped storage power station unit based on big data, which specifically includes the following steps:

[0007] S1. Setting a plurality of physical property monitoring points and a plurality of vibration signal monitoring points within a monitoring area of ​​a pumped-storage power station unit, and collecting physical property data and vibration signal data of the pumped-storage power station unit in real time based on the set physical property monitoring points and vibration signal monitoring points;

[0008] S2. Preprocessing the collected physical property data and vibration signal data using a data preprocessing method to obtain preprocessed physical property data and vibration signal data;

[0009] S3. Analyze the preprocessed physical property data and vibration signal data using a data analysis method to obtain analyzed physical property data and vibration signal data;

[0010] S31, performing modeling analysis using a three-dimensional modeling method based on the pre-processed physical property data;

[0011] S32, performing identification analysis based on the pre-processed vibration signal data using a state analysis method;

[0012] S33, summarizing the physical property data and vibration signal data after modeling analysis and identification analysis to obtain analyzed physical property data and vibration signal data;

[0013] S4. Build a fault diagnosis model based on the analyzed physical property data and vibration signal data and determine the fault level;

[0014] S5. Based on the constructed fault diagnosis model, a real-time fault prediction model is constructed through a digital twin method, and the status of the pumped storage power station units is monitored and warned in real time based on the constructed real-time fault prediction model.

[0015] Preferably, the step of setting a plurality of physical property monitoring points and a plurality of vibration signal monitoring points within a monitoring area of ​​a pumped-storage power station unit, and collecting physical property data and vibration signal data of the pumped-storage power station unit in real time based on the set physical property monitoring points and vibration signal monitoring points comprises the following steps:

[0016] S11, setting multiple physical property monitoring points and multiple vibration signal monitoring points, and collecting physical property data and vibration signal data in real time includes the following steps:

[0017] The intervals between physical property monitoring points are set based on the size of the monitoring area of ​​the pumped storage power station unit, and multiple physical property monitoring points are evenly arranged based on the set intervals;

[0018] Based on the real-time collection of the physical property data of the pumped storage power station units, the pumped storage power station units are divided into key working areas and non-key working areas;

[0019] For non-critical working areas of pumped storage power station units, four vibration signal monitoring points are set on each non-critical component, two in the horizontal direction and two in the vertical direction, and the two monitoring points in the corresponding directions are arranged at a 90-degree angle to each other;

[0020] For the key working areas of the pumped storage power station units, six vibration signal monitoring points are set on each key component, including three in the horizontal direction and three in the vertical direction, and the two monitoring points in the corresponding directions are arranged at a 60-degree angle to each other;

[0021] S12, summarizing the real-time collected physical property data and vibration signal data includes the following steps:

[0022] Number the set physical property monitoring points and vibration signal monitoring points, and construct a monitoring point matrix based on the numbers;

[0023] The monitoring point collection cycle is set, and the physical property data and vibration signal data of the pumped storage power station units collected in real time are saved based on the constructed monitoring point matrix.

[0024] Preferably, the step of preprocessing the collected physical property data and vibration signal data by a data preprocessing method to obtain the preprocessed physical property data and vibration signal data comprises the following steps:

[0025] S21. Processing the collected physical attribute data using a data preprocessing method;

[0026] Traverse each physical property monitoring point to collect the physical property data of the pumped storage power station unit in real time. When it is detected that there is missing data in the physical property data collected in real time, locate the physical property monitoring point corresponding to the missing data and perform supplementary collection;

[0027] S22. Process the collected vibration signal data through a data preprocessing method.

[0028] Preferably, the processing of the collected vibration signal data by the data preprocessing method comprises the following steps:

[0029] The vibration signal data of the pumped-storage power station unit is collected in real time by traversing each vibration signal monitoring point. When missing data is detected in the physical attribute data of the pumped-storage power station unit collected in real time, the missing data is supplemented by linear interpolation and other vibration signal monitoring points in the corresponding key components or non-key components based on the classification;

[0030] The linear interpolation calculation formula is as follows:

[0031] ;

[0032] in, represents the completed data of the jth vibration signal monitoring point, Represents the vibration signal data of the j-1th vibration signal monitoring point.

[0033] Preferably, the modeling analysis based on the pre-processed physical property data by three-dimensional modeling includes the following steps:

[0034] Assume that the collected physical property data is c, and the 3D model output by 3D modeling is ;

[0035] The mathematical expression of the support vector machine algorithm is as follows:

[0036] ;

[0037] in, represents the weight, Indicates the bias value, Represents a nonlinear mapping of low-dimensional physical property data to high-dimensional space, Represents the modeling prediction value;

[0038] Set the output 3D modeling to physical property data after analysis.

[0039] Preferably, the identification analysis based on the pre-processed vibration signal data by a state analysis method includes the following steps:

[0040] S321, mechanical vibration signal data balance analysis;

[0041] During the operation of the pumped storage power station unit, the hydraulic force will drive the blades in the pumped storage power station unit to rotate, and the rotation of the blades will generate centrifugal force; the generated centrifugal force is set to be the collected mechanical vibration signal data;

[0042] Calculate the mechanical vibration signal data collected during the operation of the pumped storage power station unit and set the threshold value of the mechanical vibration signal data;

[0043] ;

[0044] in, Indicates the mass of the rotating parts, Indicates the speed, represents the acceleration due to gravity, represents the eccentricity, Indicates the unit's rotation frequency. Represents the mechanical vibration signal data collected during the operation of the pumped storage power station unit;

[0045] S322, Hydraulic vibration signal data balance analysis;

[0046] According to the hydraulic equation, the relationship model between hydraulic vibration signal data and water flow rate is as follows:

[0047] ;

[0048] in, Indicates the pressure change during hydraulic flow. represents the resistance coefficient of the pipeline, Indicates the height difference of the water flow, represents the head vector of the water pump, It represents the function of water flow rate with respect to time;

[0049] S323 , summarizing the hydraulic vibration signal data and the mechanical vibration signal data after the balance analysis to obtain analyzed vibration signal data.

[0050] Preferably, the constructing of a fault diagnosis model based on the analyzed physical property data and vibration signal data and determining the fault level comprises the following steps:

[0051] S41. Constructing a physical fault diagnosis model based on the analyzed physical property data;

[0052] S42, determining a fault level based on the analyzed vibration signal data and the constructed physical fault diagnosis model;

[0053] The mechanical vibration signal data collected during the operation of the pumped storage power station units is calculated in real time and compared with the set mechanical vibration signal data threshold, and the faults are graded based on the comparison results and the constructed physical fault diagnosis model.

[0054] Preferably, the construction of a physical fault diagnosis model based on the analyzed physical property data comprises the following steps:

[0055] S411, reducing the dimensionality of the three-dimensional modeling in the analyzed physical property data by a dimensionality reduction method to obtain a reduced dimensional three-dimensional modeling image, and using the reduced dimensional three-dimensional modeling image as an input of a convolutional neural network;

[0056] The convolutional neural network includes: a convolutional layer, a pooling layer and a fully connected layer;

[0057] S412, based on the received 3D modeling image, extracting 3D modeling image features through a convolution operation of a convolutional layer in a convolutional neural network;

[0058] S413, processing the 3D modeling image features extracted by the convolutional layer through the activation function of the pooling layer in the convolutional neural network, iteratively convolving the input 3D modeling image through continuously stacked convolutional layers and pooling layers after the processing, and passing the features output after the convolution to the fully connected layer;

[0059] S414. Expand and combine the output features through the fully connected layer to obtain a feature vector and save it;

[0060] Compare the feature vector extracted in real time with the feature vector extracted at the previous moment, and determine whether there is a fault based on the comparison result;

[0061] Set the comparison difference threshold and calculate the comparison difference based on the data similarity formula;

[0062] A physical fault diagnosis model is constructed by combining convolutional neural networks and data similarity formula.

[0063] Preferably, the method of constructing a real-time fault prediction model based on the constructed fault diagnosis model in a digital twin manner, and performing real-time monitoring and early warning on the status of the pumped storage power station units based on the constructed real-time fault prediction model includes the following steps:

[0064] By using the control variable method, the changes in physical attribute data and vibration signal data are recorded, and the recorded changes in physical attribute data and vibration signal data are input into the constructed real-time fault prediction model to record the corresponding changes in fault levels;

[0065] Setting digital twin prediction weights based on recorded physical property data and vibration signal data changes ;

[0066] Predict the change in fault level based on the set digital twin prediction weight;

[0067] ;

[0068] in, represents the predicted change in fault level, represents the constructed fault diagnosis model;

[0069] Monitor and issue early warnings for pumped storage power station units based on real-time predicted fault level changes.

[0070] The present invention discloses a pumped storage power station unit state monitoring system based on big data, which is used to implement a pumped storage power station unit state monitoring method based on big data. The system includes: a data acquisition module, a data processing module, a data analysis module, a fault diagnosis module and a fault prediction module;

[0071] The data acquisition module is used to collect physical property data and vibration signal data of the pumped storage power station unit in real time;

[0072] The data processing module is used to process the physical property data and vibration signal data collected in real time to obtain pre-processed physical property data and vibration signal data;

[0073] The analysis module is used to analyze the pre-processed physical property data and vibration signal data;

[0074] The fault diagnosis module is used to build a fault diagnosis model based on the analysis results of the physical property data and the vibration signal data;

[0075] The fault prediction module is used to perform fault prediction in combination with a fault diagnosis model through a digital twin approach.

[0076] The beneficial effects of the present invention are:

[0077] The present invention sets multiple physical property monitoring points and multiple vibration signal monitoring points in the monitoring area of ​​the pumped-storage power station unit, and collects the physical property data and vibration signal data of the pumped-storage power station unit in real time. At the same time, the collected physical property data and vibration signal data are preprocessed by a data preprocessing method. After the preprocessing is completed, the preprocessed physical property data and vibration signal data are analyzed by a data analysis method. At the same time, a fault diagnosis model is constructed based on the analyzed physical property data and vibration signal data, and the fault level is determined. Finally, a real-time fault prediction model is constructed based on the constructed fault diagnosis model through a digital twin method, and the status of the pumped-storage power station unit is monitored and warned in real time based on the constructed real-time fault prediction model, thereby improving the accuracy of status monitoring of the water storage power station unit.

[0078] The present invention filters the collected physical property data through a data preprocessing method, and processes the vibration signal data by means of data filtering and filling in missing data, thereby improving the reliability of data processing.

[0079] The present invention performs modeling based on preprocessed physical property data through a three-dimensional modeling method, and simultaneously identifies and analyzes the processed vibration signal data through a state analysis method, and ensures the accuracy of data analysis based on mechanical vibration signal data balance analysis and hydraulic vibration signal data balance analysis.

[0080] The present invention constructs a physical fault diagnosis model based on the analyzed physical property data. Specifically, it determines whether a physical fault exists through feature recognition and feature comparison. After the physical fault diagnosis model is constructed, the fault level is determined in combination with the analyzed vibration signal data, thereby improving the rationality of fault analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in 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 only 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.

[0082] Figure 1 The figure is a flow chart of the method for monitoring the status of a pumped storage power station unit based on big data according to the present invention. DETAILED DESCRIPTION

[0083] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0084] In a specific embodiment of the present invention,

[0085] Reference Figure 1 As shown, the present invention provides a method for monitoring the status of a pumped storage power station unit based on big data, which specifically includes the following steps:

[0086] S1. Setting a plurality of physical property monitoring points and a plurality of vibration signal monitoring points within a monitoring area of ​​a pumped-storage power station unit, and collecting physical property data and vibration signal data of the pumped-storage power station unit in real time based on the set physical property monitoring points and vibration signal monitoring points;

[0087] S2. Preprocessing the collected physical property data and vibration signal data using a data preprocessing method to obtain preprocessed physical property data and vibration signal data;

[0088] S3. Analyze the preprocessed physical property data and vibration signal data using a data analysis method to obtain analyzed physical property data and vibration signal data;

[0089] S31, performing modeling analysis using a three-dimensional modeling method based on the pre-processed physical property data;

[0090] S32, performing identification analysis based on the pre-processed vibration signal data using a state analysis method;

[0091] S33, summarizing the physical property data and vibration signal data after modeling analysis and identification analysis to obtain analyzed physical property data and vibration signal data;

[0092] S4. Build a fault diagnosis model based on the analyzed physical property data and vibration signal data and determine the fault level;

[0093] S5. Based on the constructed fault diagnosis model, a real-time fault prediction model is constructed through a digital twin approach, and the status of the pumped storage power station units is monitored and warned in real time based on the constructed real-time fault prediction model;

[0094] Further, refer to Figure 1As shown, a plurality of physical property monitoring points and a plurality of vibration signal monitoring points are arranged in the pumped storage power station unit monitoring area, and the physical property data and vibration signal data of the pumped storage power station unit are collected in real time based on the arranged physical property monitoring points and vibration signal monitoring points, including the following steps:

[0095] S11, a plurality of physical property monitoring points and a plurality of vibration signal monitoring points are arranged, and physical property data and vibration signal data are collected in real time, including the following steps:

[0096] The interval between the physical property monitoring points is set based on the size of the pumped storage power station unit monitoring area, and a plurality of physical property monitoring points are evenly arranged based on the set interval;

[0097] Further, the physical property data of the pumped storage power station unit is collected in real time based on the plurality of arranged physical property monitoring points;

[0098] The physical property data includes specific parameters of each key structural part of the pumped storage power station unit;

[0099] Further, the pumped storage power station unit is divided into a key working area and a non-key working area based on the real-time collection of the physical property data of the pumped storage power station unit;

[0100] For the non-key working area of the pumped storage power station unit, 4 vibration signal monitoring points are set on each non-key component, of which 2 are in the horizontal direction and 2 are in the vertical direction, and the two monitoring points in the corresponding direction are arranged at an angle of 90 degrees;

[0101] For the key working area of the pumped storage power station unit, 6 vibration signal monitoring points are set on each key component, of which 3 are in the horizontal direction and 3 are in the vertical direction, and the two monitoring points in the corresponding direction are arranged at an angle of 60 degrees;

[0102] Further, the vibration signal data of the pumped storage power station unit is collected in real time based on the plurality of arranged vibration signal monitoring points;

[0103] The vibration signal data includes mechanical vibration signal and hydraulic vibration signal;

[0104] S12, the collected physical property data and vibration signal data are summarized, including the following steps:

[0105] The physical property monitoring points and the vibration signal monitoring points are numbered, and a monitoring point matrix is constructed based on the numbering;

[0106] Further, a monitoring point collection cycle is set, and the physical property data and vibration signal data of the pumped storage power station unit collected in real time are saved based on the constructed monitoring point matrix;

[0107] Further, refer to Figure 1 As shown, preprocessing the collected physical property data and vibration signal data by a data preprocessing method to obtain the preprocessed physical property data and vibration signal data includes the following steps:

[0108] S21. Processing the collected physical attribute data using a data preprocessing method;

[0109] Traverse each physical property monitoring point to collect the physical property data of the pumped storage power station unit in real time. When it is detected that there is missing data in the physical property data collected in real time, locate the physical property monitoring point corresponding to the missing data and perform supplementary collection;

[0110] S22, processing the collected vibration signal data through a data preprocessing method;

[0111] The vibration signal data of the pumped-storage power station unit is collected in real time by traversing each vibration signal monitoring point. When missing data is detected in the physical attribute data of the pumped-storage power station unit collected in real time, the missing data is supplemented by linear interpolation and other vibration signal monitoring points in the corresponding key components or non-key components based on the classification;

[0112] The linear interpolation calculation formula is as follows:

[0113] ;

[0114] in, represents the completed data of the jth vibration signal monitoring point, Represents the vibration signal data of the j-1th vibration signal monitoring point;

[0115] Further, refer to Figure 1 As shown, modeling and analysis based on the pre-processed physical property data through three-dimensional modeling includes the following steps:

[0116] Assume that the collected physical property data is c, and the 3D model output by 3D modeling is ;

[0117] The mathematical expression of the support vector machine algorithm is as follows:

[0118] ;

[0119] in, represents the weight, Indicates the bias value, Represents a nonlinear mapping of low-dimensional physical property data to high-dimensional space, Represents the modeling prediction value;

[0120] Furthermore, the output three-dimensional modeling is set as the physical property data after analysis;

[0121] Further, refer to Figure 1 As shown, the identification analysis based on the pre-processed vibration signal data by the state analysis method includes the following steps:

[0122] S321, mechanical vibration signal data balance analysis;

[0123] During the operation of the pumped storage power station unit, the hydraulic force will drive the blades in the pumped storage power station unit to rotate, and the rotation of the blades will generate centrifugal force; the generated centrifugal force is set to be the collected mechanical vibration signal data;

[0124] Furthermore, the mechanical vibration signal data collected during the operation of the pumped storage power station unit is calculated, and a threshold value of the mechanical vibration signal data is set;

[0125] ;

[0126] in, Indicates the mass of the rotating parts, Indicates the speed, represents the acceleration due to gravity, represents the eccentricity, Indicates the unit's rotation frequency. Represents the mechanical vibration signal data collected during the operation of the pumped storage power station unit;

[0127] S322, Hydraulic vibration signal data balance analysis;

[0128] According to the hydraulic equation, the relationship model between hydraulic vibration signal data and water flow rate is as follows:

[0129] ;

[0130] in, Indicates the pressure change during hydraulic flow. represents the resistance coefficient of the pipeline, Indicates the height difference of the water flow, represents the head vector of the water pump, It represents the function of water flow rate with respect to time;

[0131] S323, summarizing the hydraulic vibration signal data and the mechanical vibration signal data after the balance analysis to obtain analyzed vibration signal data;

[0132] Further, refer to Figure 1 As shown, building a fault diagnosis model based on the analyzed physical property data and vibration signal data and determining the fault level includes the following steps:

[0133] S41. Constructing a physical fault diagnosis model based on the analyzed physical property data;

[0134] S411, reducing the dimensionality of the three-dimensional modeling in the analyzed physical property data by a dimensionality reduction method to obtain a reduced dimensional three-dimensional modeling image, and using the reduced dimensional three-dimensional modeling image as an input of a convolutional neural network;

[0135] The convolutional neural network includes: a convolutional layer, a pooling layer and a fully connected layer;

[0136] S412, based on the received 3D modeling image, extracting 3D modeling image features through a convolution operation of a convolutional layer in a convolutional neural network;

[0137] The convolution calculation formula is as follows:

[0138] ;

[0139] in, represents the input 3D modeling image, represents the weight of the corresponding convolution kernel, b represents the bias value, represents the output features;

[0140] S413, processing the 3D modeling image features extracted by the convolutional layer through the activation function of the pooling layer in the convolutional neural network, iteratively convolving the input 3D modeling image through continuously stacked convolutional layers and pooling layers after the processing, and passing the features output after the convolution to the fully connected layer;

[0141] S414. Expand and combine the output features through the fully connected layer to obtain a feature vector and save it;

[0142] Furthermore, the feature vector extracted in real time is compared with the feature vector extracted at the previous moment, and whether there is a fault is determined based on the comparison result;

[0143] Set the comparison difference threshold and calculate the comparison difference based on the data similarity formula;

[0144] The data similarity formula is as follows:

[0145] ;

[0146] in, Represents the feature vector and eigenvectors The similarity value between Indicates the real-time extraction of feature vectors of 3D modeling images, Represents the feature vector of the 3D modeling image extracted at the last moment;

[0147] Furthermore, a physical fault diagnosis model is constructed by combining convolutional neural networks and data similarity formula;

[0148] S42. Determine the fault level based on the analyzed vibration signal data and the constructed physical fault diagnosis model;

[0149] The mechanical vibration signal data collected during the operation of the pumped storage power station units is calculated in real time and compared with the set mechanical vibration signal data threshold. The faults are then classified based on the comparison results and the constructed physical fault diagnosis model.

[0150] Level 1 fault: The real-time collected mechanical vibration signal data exceeds the set mechanical vibration signal data threshold by 5% but less than 10%, and the difference between the real-time extracted feature vector and the feature vector extracted at the previous moment exceeds 5% but less than 10%;

[0151] Level 2 fault: The real-time collected mechanical vibration signal data exceeds 10% of the set mechanical vibration signal data threshold, or the difference between the real-time extracted feature vector and the feature vector extracted at the previous moment exceeds 10%;

[0152] Level 3 fault: The real-time collected mechanical vibration signal data exceeds 10% of the set mechanical vibration signal data threshold, and the difference between the real-time extracted feature vector and the feature vector extracted at the previous moment exceeds 10%;

[0153] Further, refer to Figure 1 As shown in FIG, a real-time fault prediction model is constructed based on the constructed fault diagnosis model through a digital twin method, and the real-time monitoring and early warning of the status of the pumped storage power station unit based on the constructed real-time fault prediction model include the following steps:

[0154] By using the control variable method, the changes in physical attribute data and vibration signal data are recorded, and the recorded changes in physical attribute data and vibration signal data are input into the constructed real-time fault prediction model to record the corresponding changes in fault levels;

[0155] Setting digital twin prediction weights based on recorded physical property data and vibration signal data changes ;

[0156] Predict the change in fault level based on the set digital twin prediction weight;

[0157] ;

[0158] in, represents the predicted change in fault level, represents the constructed fault diagnosis model;

[0159] Monitor and provide early warning for pumped storage power station units based on real-time predicted fault level changes;

[0160] In a specific embodiment, the pumped storage power station unit state monitoring system based on big data is used to implement a pumped storage power station unit state monitoring method based on big data, and the system includes: a data acquisition module, a data processing module, a data analysis module, a fault diagnosis module, and a fault prediction module;

[0161] The data acquisition module is used to collect physical property data and vibration signal data of the pumped storage power station unit in real time;

[0162] The data processing module is used to process the physical property data and vibration signal data collected in real time to obtain pre-processed physical property data and vibration signal data;

[0163] The analysis module is used to analyze the pre-processed physical property data and vibration signal data;

[0164] The fault diagnosis module is used to build a fault diagnosis model based on the analysis results of the physical property data and the vibration signal data;

[0165] The fault prediction module is used to perform fault prediction in combination with a fault diagnosis model through a digital twin approach.

[0166] It should be noted that

[0167] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they should all fall within the scope of protection of the present invention.

Claims

1. A method for monitoring the status of a pumped storage power station unit based on big data, characterized in that: The following steps are involved: S1. Setting a plurality of physical property monitoring points and a plurality of vibration signal monitoring points within a monitoring area of ​​a pumped-storage power station unit, and collecting physical property data and vibration signal data of the pumped-storage power station unit in real time based on the set physical property monitoring points and vibration signal monitoring points; S2. Preprocessing the collected physical property data and vibration signal data using a data preprocessing method to obtain preprocessed physical property data and vibration signal data; S3. Analyze the preprocessed physical property data and vibration signal data using a data analysis method to obtain analyzed physical property data and vibration signal data; S31, performing modeling analysis using a three-dimensional modeling method based on the pre-processed physical property data; S32, performing identification analysis based on the pre-processed vibration signal data using a state analysis method; S321, mechanical vibration signal data balance analysis; During the operation of the pumped storage power station unit, the hydraulic force will drive the blades in the pumped storage power station unit to rotate, and the rotation of the blades will generate centrifugal force; the generated centrifugal force is set to be the collected mechanical vibration signal data; Calculate the mechanical vibration signal data collected during the operation of the pumped storage power station unit and set the threshold value of the mechanical vibration signal data; ； in, represents the mass of the rotating parts, Indicates the speed, represents the acceleration due to gravity, represents the eccentricity, Indicates the unit's rotation frequency. Represents the mechanical vibration signal data collected during the operation of the pumped storage power station unit; S322, Hydraulic vibration signal data balance analysis; According to the hydraulic equation, the relationship model between hydraulic vibration signal data and water flow rate is as follows: ； in, Indicates the pressure change during hydraulic flow. represents the resistance coefficient of the pipeline, Indicates the height difference of the water flow, represents the head vector of the water pump, It represents the function of water flow rate with respect to time; S323, summarizing the hydraulic vibration signal data and the mechanical vibration signal data after the balance analysis to obtain analyzed vibration signal data; S33, summarizing the physical property data and vibration signal data after modeling analysis and identification analysis to obtain analyzed physical property data and vibration signal data; S4. Build a fault diagnosis model based on the analyzed physical property data and vibration signal data and determine the fault level; S5. Based on the constructed fault diagnosis model, a real-time fault prediction model is constructed through a digital twin method, and the status of the pumped storage power station units is monitored and warned in real time based on the constructed real-time fault prediction model.

2. A method for monitoring the status of a pumped storage power station unit based on big data according to claim 1, characterized in that: The method of setting a plurality of physical property monitoring points and a plurality of vibration signal monitoring points within the monitoring area of ​​the pumped storage power station unit, and collecting the physical property data and vibration signal data of the pumped storage power station unit in real time based on the set physical property monitoring points and vibration signal monitoring points comprises the following steps: S11, setting multiple physical property monitoring points and multiple vibration signal monitoring points, and collecting physical property data and vibration signal data in real time includes the following steps: The intervals between physical property monitoring points are set based on the size of the monitoring area of ​​the pumped storage power station unit, and multiple physical property monitoring points are evenly arranged based on the set intervals; Based on the real-time collection of the physical property data of the pumped storage power station units, the pumped storage power station units are divided into key working areas and non-key working areas; For non-critical working areas of pumped storage power station units, four vibration signal monitoring points are set on each non-critical component, two in the horizontal direction and two in the vertical direction, and the two monitoring points in the corresponding directions are arranged at a 90-degree angle to each other; For the key working areas of the pumped storage power station units, six vibration signal monitoring points are set on each key component, including three in the horizontal direction and three in the vertical direction, and the two monitoring points in the corresponding directions are arranged at a 60-degree angle to each other; S12, summarizing the real-time collected physical property data and vibration signal data includes the following steps: Number the set physical property monitoring points and vibration signal monitoring points, and construct a monitoring point matrix based on the numbers; The monitoring point collection cycle is set, and the physical property data and vibration signal data of the pumped storage power station units collected in real time are saved based on the constructed monitoring point matrix.

3. The method for monitoring the status of a pumped storage power station unit based on big data according to claim 1, characterized in that: The method of preprocessing the collected physical property data and vibration signal data by a data preprocessing method to obtain the preprocessed physical property data and vibration signal data includes the following steps: S21. Processing the collected physical attribute data using a data preprocessing method; Traverse each physical property monitoring point to collect the physical property data of the pumped storage power station unit in real time. When it is detected that there is missing data in the physical property data collected in real time, locate the physical property monitoring point corresponding to the missing data and perform supplementary collection; S22. Process the collected vibration signal data through a data preprocessing method.

4. A method for monitoring the status of a pumped storage power station unit based on big data according to claim 3, characterized in that: The processing of the collected vibration signal data by the data preprocessing method comprises the following steps: The vibration signal data of the pumped-storage power station unit is collected in real time by traversing each vibration signal monitoring point. When missing data is detected in the physical attribute data of the pumped-storage power station unit collected in real time, the missing data is supplemented by linear interpolation and other vibration signal monitoring points in the corresponding key components or non-key components based on the classification; The linear interpolation calculation formula is as follows: ； in, represents the completed data of the jth vibration signal monitoring point, Represents the vibration signal data of the j-1th vibration signal monitoring point.

5. The method for monitoring the status of a pumped storage power station unit based on big data according to claim 1, characterized in that: The modeling analysis based on the pre-processed physical property data by three-dimensional modeling includes the following steps: Assume that the collected physical property data is c, and the 3D model output by 3D modeling is ; The mathematical expression of the support vector machine algorithm is as follows: ； in, represents the weight, Indicates the bias value, Represents a nonlinear mapping of low-dimensional physical property data to high-dimensional space, Represents the modeling prediction value; Set the output 3D modeling to physical property data after analysis.

6. The method for monitoring the status of a pumped storage power station unit based on big data according to claim 1, characterized in that: The method of constructing a fault diagnosis model based on the analyzed physical property data and vibration signal data and determining the fault level includes the following steps: S41. Constructing a physical fault diagnosis model based on the analyzed physical property data; S42. Determine the fault level based on the analyzed vibration signal data and the constructed physical fault diagnosis model; The mechanical vibration signal data collected during the operation of the pumped storage power station units is calculated in real time and compared with the set mechanical vibration signal data threshold, and the faults are graded based on the comparison results and the constructed physical fault diagnosis model.

7. A method for monitoring the status of a pumped storage power station unit based on big data according to claim 6, characterized in that: The construction of a physical fault diagnosis model based on the analyzed physical property data comprises the following steps: S411, reducing the dimensionality of the three-dimensional modeling in the analyzed physical property data by a dimensionality reduction method to obtain a reduced dimensional three-dimensional modeling image, and using the reduced dimensional three-dimensional modeling image as an input of a convolutional neural network; The convolutional neural network includes: a convolutional layer, a pooling layer and a fully connected layer; S412, based on the received 3D modeling image, extracting 3D modeling image features through a convolution operation of a convolutional layer in a convolutional neural network; S413, processing the 3D modeling image features extracted by the convolutional layer through the activation function of the pooling layer in the convolutional neural network, iteratively convolving the input 3D modeling image through continuously stacked convolutional layers and pooling layers after the processing, and passing the features output after the convolution to the fully connected layer; S414. Expand and combine the output features through the fully connected layer to obtain a feature vector and save it; Compare the feature vector extracted in real time with the feature vector extracted at the previous moment, and determine whether there is a fault based on the comparison result; Set the comparison difference threshold and calculate the comparison difference based on the data similarity formula; A physical fault diagnosis model is constructed by combining convolutional neural networks and data similarity formula.

8. The method for monitoring the status of a pumped storage power station unit based on big data according to claim 1, characterized in that: The method of constructing a real-time fault prediction model based on the constructed fault diagnosis model in a digital twin manner, and performing real-time monitoring and early warning of the status of the pumped storage power station unit based on the constructed real-time fault prediction model includes the following steps: By using the control variable method, the changes in physical attribute data and vibration signal data are recorded, and the recorded changes in physical attribute data and vibration signal data are input into the constructed real-time fault prediction model to record the corresponding changes in fault levels; Setting digital twin prediction weights based on recorded physical property data and vibration signal data changes ; Predict the change in fault level based on the set digital twin prediction weight; ； in, represents the predicted change in fault level, represents the constructed fault diagnosis model; Monitor and issue early warnings for pumped storage power station units based on real-time predicted fault level changes.

9. A system for implementing the method for monitoring the status of a pumped storage power station unit based on big data according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, data processing module, data analysis module, fault diagnosis module and fault prediction module; The data acquisition module is used to collect physical property data and vibration signal data of the pumped storage power station unit in real time; The data processing module is used to process the physical property data and vibration signal data collected in real time to obtain pre-processed physical property data and vibration signal data; The analysis module is used to analyze the pre-processed physical property data and vibration signal data; The fault diagnosis module is used to build a fault diagnosis model based on the analysis results of the physical property data and the vibration signal data; The fault prediction module is used to perform fault prediction in combination with a fault diagnosis model through a digital twin approach.

Images