A method and system for automatically storing digital power plant data

Abstract

The present invention relates to the technical field of data storage, and in particular to an automatic storage method and system for digital power plant data, comprising the following steps: S1, data acquisition: real-time acquisition of digital power plant data; S2, data preprocessing: preprocessing the acquired digital power plant data; S3, context-related storage: context-related analysis of the preprocessed digital power plant data; S4, data compression and encryption: compression and encryption of the associated stored digital power plant data; S5, data synchronization and backup: regular synchronization and backup of the digital power plant data stored in a database; S6, data retrieval and analysis: rapid retrieval of context data related to the target digital power plant data. The present invention improves the storage efficiency and intelligence level of data, and provides more accurate support for the operation optimization and fault diagnosis of power plants.

Description

Technical Field

[0001] The present invention relates to the technical field of data storage, and in particular to an automatic storage method and system for digital power plant data. Background Art

[0002] With the rapid development of the energy industry, digital power plants, as an important part of smart grids, have gradually become a key technical means to improve power production efficiency and safety management. Digital power plants deploy a large number of sensors and intelligent devices to monitor and collect various data generated during the operation of power plants in real time, such as electrical parameters, environmental conditions, equipment status and production processes. The efficient storage and intelligent analysis of these data are crucial to optimizing power plant operations, preventing failures and improving the stability of the overall power system. However, with the continuous increase in the amount of data, how to efficiently manage and utilize these data has become an important challenge facing digital power plants.

[0003] In the prior art, the data storage and management methods of digital power plants usually lack the ability to intelligently analyze and dynamically optimize data associations, resulting in low data storage efficiency and difficulty in ensuring the real-time and consistency of data under complex operating environments. In addition, the security and data recovery capabilities of the prior art also have certain deficiencies. The protection measures for key data are relatively simple and cannot fully meet the modern digital power plant's needs for efficient, secure, and intelligent data management, especially in terms of data compression and encryption, dynamic data backup and redundancy management, and contextual data retrieval and analysis. Existing solutions are difficult to cope with the complexity and variability in actual applications, and there is a lot of room for improvement.

[0004] The purpose of the present invention is to provide a method and system for automatically storing digital power plant data, so as to provide a more efficient, reliable and intelligent solution for the data management of the digital power plant. Summary of the invention

[0005] The invention provides a method and system for automatically storing digital power plant data.

[0006] A method for automatically storing digital power plant data comprises the following steps:

[0007] S1, data collection: Use multiple sensors deployed in the digital power plant to collect digital power plant data in real time, including electrical parameter data (voltage, current, frequency, power), environmental parameter data (temperature, pressure, humidity), equipment status data (vibration, speed, oil level and oil temperature), safety monitoring data (smoke composition, fire alarm), production process data (fuel consumption, water treatment data) and real-time operation data (production plan execution, equipment maintenance status);

[0008] S2, data preprocessing: preprocessing the collected digital power plant data, including data denoising, missing value filling and data format unification;

[0009] S3, context-related storage: Use graph neural networks (GNNs) to perform context-related analysis on pre-processed digital power plant data, including:

[0010] S31, Data Relationship Modeling: Use graph neural networks (GNNs) to model the time series and spatial position relationships of digital power plant data from different sensors, and build graph structures between data to identify potential associations between different data sources;

[0011] S32, dynamic correlation optimization: According to the real-time changes in the operating status of the power plant, the correlation between the digital power plant data is dynamically adjusted through the graph neural network (GNN);

[0012] S33, associated data storage: bundling and storing the associated optimized digital power plant data;

[0013] S4, data compression and encryption: compressing and encrypting the associated stored digital power plant data;

[0014] S5, data synchronization and backup: regularly synchronize and back up the digital power plant data stored in the database. During the synchronization and backup process, an adaptive redundancy strategy optimization algorithm is used to dynamically adjust the redundancy level according to the importance and access frequency of the data;

[0015] S6, data retrieval and analysis: By establishing a graph-based indexing mechanism, contextual data related to the target digital power plant data can be quickly retrieved and analyzed to support power plant operation optimization and fault diagnosis.

[0016] Optionally, the data collection in S1 includes:

[0017] S11, electrical parameter data collection: deploy current sensors, voltage sensors and frequency sensors in power generation equipment and distribution facilities to collect electrical parameter data in real time;

[0018] S12, Environmental parameter data collection: Deploy temperature sensors, pressure sensors, and humidity sensors in boilers, pipelines, and cooling facilities to monitor environmental parameters;

[0019] S13, equipment status data collection: Install vibration sensors, speed sensors, oil level sensors and oil temperature sensors on mechanical equipment (turbines, generators, pumps) to obtain equipment status data;

[0020] S14, safety monitoring data collection: smoke composition sensors and fire alarms are configured in emission monitoring facilities and fire safety units to collect safety monitoring data;

[0021] S15, production process data collection: fuel consumption sensors and water quality monitoring sensors are set in the fuel supply and water treatment units to collect production process data;

[0022] S16, real-time operation data collection: Deploy relevant sensors and data interfaces in the power plant management unit to collect production plan execution and equipment maintenance status in real time.

[0023] Optionally, the data preprocessing in S2 includes:

[0024] S21, data denoising: wavelet transform algorithm is used to perform multi-scale analysis on digital power plant data, decompose the data into different frequency bands, and perform threshold processing on the high-frequency part to remove noise;

[0025] S22, missing value filling: For the missing values ​​in the digital power plant data, the K nearest neighbor algorithm is used to fill the missing values. The nearest data point is selected by calculating the Euclidean distance between the data point and its surrounding known data points. The data points are weighted averaged;

[0026] Data format unification: Regular expression matching and conversion algorithms are used to standardize the data format into a unified timestamp, unit, and data type.

[0027] Optionally, the data relationship modeling in S31 includes:

[0028] S311, Graph Structure Construction: Use Graph Neural Network (GNN) to model the time series and spatial position relationship of digital power plant data from different sensors, and regard each sensor data as a node in the graph , the edges between nodes Represents the association between sensor data and the weight of the edge The calculation is based on the similarity of the time series and the proximity of the spatial position, expressed as:

[0029] ;

[0030] in, For sensor and The similarity in time series, For sensor and The distance in space, and To adjust the hyperparameters of weights;

[0031] S312, node feature initialization: each node The initial eigenvector of It is obtained by extracting the features of the corresponding sensor data, including the statistical features of the sensor measurement values ​​and the sensor location and type, and is expressed as:

[0032] ;

[0033] in, and Respectively represent sensors The mean and standard deviation of the collected data, Indicates sensor spatial location, Indicates the type of sensor;

[0034] S313, graph convolution operation: feature propagation and update of nodes in the graph structure through the graph convolutional neural network (GCN), expressed as:

[0035] ;

[0036] in, Indicates Node in layer The characteristic vector of For Node The set of neighbor nodes of For the The weight matrix of the layer, is the ReLU activation function, and For Node and The degree of

[0037] S314, Potential Association Identification: Through layer-by-layer feature updates of multi-layer graph convolutional neural networks (GCN), a high-dimensional feature representation of each node is finally obtained. , combined with the weights of the edges in the graph Identify potential correlations between sensor data, expressed as:

[0038] ;

[0039] in, is the final incidence matrix, is the sigmoid activation function, and They are respectively Post-layer node and nodes The final feature vector of .

[0040] Optionally, the dynamic association optimization in S32 includes:

[0041] S321, Real-time status monitoring and input update: Real-time monitoring of the operating status parameters of the power plant, including load changes, environmental conditions, and equipment status, and using the operating status parameters as dynamic inputs to the graph neural network (GNN) to update the feature representation of each node , expressed as:

[0042] ;

[0043] in, For Node In the The feature vector after layer graph convolution, is the ReLU activation function, For Node The set of neighbor nodes of For Node and nodes The edge weights between is the mapping function, according to the current power plant operation status Adjust edge weights , and Node and nodes The degree of For the The weight matrix of the layer, Neighbor Node In the Feature vector after layer graph convolution;

[0044] S322, dynamic adjustment of edge weights: adjust the node weights according to the real-time changes in the power plant operation status. and nodes The edge weight between , so that the association relationship between nodes dynamically reflects the current operating conditions of the power plant, expressed as:

[0045] ;

[0046] in, is the initial edge weight;

[0047] S323, Multi-layer Graph Convolutional Network Optimization: After iterative updates of multi-layer graph convolutional networks (GCN), node feature representation The edge weights between nodes Dynamically reflects the optimal correlation relationship under the current power plant operation status, and outputs node features and edge weights to construct an optimized correlation matrix , expressed as:

[0048] ;

[0049] in, is the sigmoid activation function, For Node In the The final feature vector after layer graph convolution, For Node In the The final feature vector after the layer graph convolution.

[0050] Optionally, the associated data storage in S33 includes:

[0051] S331, Data bundling strategy generation: Based on the association matrix optimized by graph neural network (GNN) , determine the set of data nodes with strong associations , if the optimized correlation matrix The value exceeds the preset threshold , then the corresponding node and nodes Join the same collection ;

[0052] S332, data block division and bundled storage: each associated data set The digital power plant data after correlation optimization is packaged into independent data blocks , and separate data blocks Bundle storage.

[0053] Optionally, the data compression and encryption in S4 includes:

[0054] S41, data compression: use a hybrid compression algorithm to compress the associated stored digital power plant data, use Huffman coding to compress the real-time digital power plant data, and use discrete wavelet transform to compress the historical digital power plant data;

[0055] S42, Data Encryption: The compressed digital power plant data is encrypted using the AES algorithm. The encryption process includes generating a symmetric key and Block encryption is performed, expressed as:

[0056] ;

[0057] in, is the encrypted data block, is the compressed data block, is the symmetric encryption key, is the encryption function.

[0058] Optionally, the adaptive redundancy strategy optimization algorithm in S5 includes:

[0059] S51, data importance weight calculation: for each type of digital power plant data , calculate its importance weight in power plant operation , expressed as:

[0060] ;

[0061] in, To meet the real-time data requirements, Because of the criticality of the data, and is the weight coefficient;

[0062] S52, data access frequency calculation: by recording the data access log, calculate each data block Frequency of visits ;

[0063] S53, Dynamic adjustment of redundancy level: based on the importance weight of the data and access frequency , dynamically adjust each data block Redundancy level , expressed as:

[0064] ;

[0065] in, For data blocks The number of redundant copies of is the adjustment factor, is the number of all data blocks;

[0066] S54, Dynamic Allocation and Storage: Based on Calculated Redundancy Level , for data blocks Dynamically allocate the corresponding number of redundant copies and distribute the redundant copies to different storage nodes.

[0067] Optionally, the data retrieval and analysis in S6 includes:

[0068] S61, Establishment of an indexing mechanism based on a graph structure: By analyzing the graph structure of digital power plant data, a multi-level indexing mechanism is constructed, including:

[0069] S611, global index: partition the entire graph structure, each partition contains an associated set of nodes, which is used to locate the partition where the target node is located;

[0070] S612, local index: in each partition, based on the edge weights between nodes, establish a local index for retrieving context data associated with the target node;

[0071] S62, fast data retrieval algorithm: using the index mechanism based on the graph structure, locate the partition where the target data node is located through the global index, and retrieve the context data closely related to the target node through the local index, expressed as:

[0072] ;

[0073] in, Representation Node A collection of context data, Representation Node and nodes The edge weights between is the preset edge weight threshold, Representation Node The set of neighbor nodes of

[0074] S63, context data analysis: Analyze the retrieved context data, use graph neural network (GNN) to mine the complex correlation between data, and combine the operation status of the power plant to optimize the operation and diagnose faults of the power plant, including:

[0075] S631, feature extraction: extract features from context data, including time series features, equipment operation status features, data association features, failure mode features, spatial and topological features, and access frequency features, and generate context feature vectors , expressed as:

[0076] ;

[0077] in, For Node and nodes The edge weights between The context node The eigenvector of

[0078] S632, relational reasoning: Use graph neural network to reason about context feature vectors and derive the correlation and fault cause with target data nodes, expressed as:

[0079] ;

[0080] in, For Node and nodes The inference relationship score between and Respectively represent nodes and nodes go through The final feature vector after layer graph convolution, is the weight matrix, is the bias term, is the sigmoid activation function.

[0081] An automatic storage system for digital power plant data, used to implement the above-mentioned automatic storage method for digital power plant data, includes the following modules:

[0082] Data acquisition module: uses multiple sensors deployed in the digital power plant to collect digital power plant data in real time, including electrical parameter data, environmental parameter data, equipment status data, safety monitoring data, production process data and real-time operation data;

[0083] Data preprocessing module: preprocess the collected digital power plant data, including data denoising, missing value filling and data format unification;

[0084] Context-associated storage module: Use graph neural network to perform context-associated analysis on pre-processed digital power plant data and perform associative storage;

[0085] Data compression and encryption module: compress and encrypt the associated stored digital power plant data;

[0086] Data synchronization and backup module: regularly synchronize and back up the digital power plant data stored in the database. Adopt an adaptive redundancy strategy optimization algorithm during the synchronization and backup process to dynamically adjust the redundancy level according to the importance and access frequency of the data.

[0087] Data retrieval and analysis module: By establishing a graph-based indexing mechanism, contextual data related to the target digital power plant data can be quickly retrieved and analyzed to support power plant operation optimization and fault diagnosis.

[0088] Beneficial effects of the present invention:

[0089] The present invention, through the deployment of multiple sensors in the digital power plant, collects real-time data covering electrical parameters, environmental parameters, equipment status, safety monitoring, production process and real-time operation data, ensuring the comprehensiveness and accuracy of data collection. At the same time, through data denoising, missing value filling and data format unification, the consistency and reliability of the data are greatly improved, providing a solid foundation for subsequent data storage, analysis and decision support.

[0090] The present invention, by utilizing graph neural networks for contextual association storage, can capture the complex spatiotemporal relationships in digital power plant data, realize data relationship modeling, dynamic association optimization, and bundled storage of associated data, and can flexibly respond to real-time changes in the operating status of the power plant by dynamically adjusting edge weights and utilizing multi-layer graph convolutional networks to optimize data association, thereby ensuring the accuracy and consistency of association relationships during data storage and analysis, improving data storage efficiency and intelligence, and providing more accurate support for power plant operation optimization and fault diagnosis.

[0091] The present invention reduces the space occupied by data storage and ensures the security of data during transmission and storage by adopting a hybrid compression algorithm and an advanced encryption standard algorithm. At the same time, the redundancy level of the data is dynamically adjusted through an adaptive redundancy strategy optimization algorithm to ensure that key data has sufficient redundant protection during high-frequency access and fault recovery. Finally, the indexing mechanism and fast data retrieval algorithm based on the graph structure, combined with the in-depth analysis of contextual data, improves the speed and accuracy of data retrieval, effectively supports the intelligent operation and preventive maintenance of power plants, and significantly improves the operating efficiency and overall safety of power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0093] Figure 1 A schematic diagram of a storage method flow chart of an embodiment of the present invention;

[0094] Figure 2 Schematic diagram of system function modules according to an embodiment of the present invention. DETAILED DESCRIPTION

[0095] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0096] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0097] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0098] like Figure 1 As shown, a method for automatically storing digital power plant data includes the following steps:

[0099] S1, data collection: Use multiple sensors deployed in the digital power plant to collect digital power plant data in real time, including electrical parameter data (voltage, current, frequency, power), environmental parameter data (temperature, pressure, humidity), equipment status data (vibration, speed, oil level and oil temperature), safety monitoring data (smoke composition, fire alarm), production process data (fuel consumption, water treatment data) and real-time operation data (production plan execution, equipment maintenance status);

[0100] S2, data preprocessing: preprocessing the collected digital power plant data, including data denoising, missing value filling and data format unification to ensure data consistency and reliability;

[0101] S3, context-related storage: Use graph neural networks (GNNs) to perform context-related analysis on pre-processed digital power plant data, including:

[0102] S31, Data Relationship Modeling: Use graph neural networks (GNNs) to model the time series and spatial position relationships of digital power plant data from different sensors, and build graph structures between data to identify potential associations between different data sources;

[0103] S32, dynamic association optimization: According to the real-time changes in the operating status of the power plant, the association between the digital power plant data is dynamically adjusted through the graph neural network (GNN) to ensure that the data classification and storage can reflect the data association under the current system status;

[0104] S33, associated data storage: Bundle and store the associated optimized digital power plant data to ensure that these associated data can be quickly and accurately extracted and used in the subsequent retrieval and analysis process;

[0105] S4, data compression and encryption: compressing and encrypting the associated stored digital power plant data;

[0106] S5, data synchronization and backup: regularly synchronize and back up the digital power plant data stored in the database to ensure the integrity and recoverability of the data. During the synchronization and backup process, an adaptive redundancy strategy optimization algorithm is used to dynamically adjust the redundancy level according to the importance and access frequency of the data to optimize the use of storage resources;

[0107] S6, data retrieval and analysis: by establishing a graph-based indexing mechanism, contextual data related to the target digital power plant data can be quickly retrieved and analyzed to support power plant operation optimization and fault diagnosis;

[0108] Through the above steps, not only the efficiency and security of data storage are improved, but also the relevance and availability of data in a dynamic environment are ensured, providing more intelligent and accurate support for the operation optimization and fault diagnosis of digital power plants.

[0109] Data collection in S1 includes:

[0110] S11, electrical parameter data collection: deploy current sensors, voltage sensors and frequency sensors in power generation equipment and distribution facilities to collect electrical parameter data in real time;

[0111] S12, Environmental parameter data collection: Deploy temperature sensors, pressure sensors, and humidity sensors in boilers, pipelines, and cooling facilities to monitor environmental parameters;

[0112] S13, equipment status data collection: Install vibration sensors, speed sensors, oil level sensors and oil temperature sensors on mechanical equipment (turbines, generators, pumps) to obtain equipment status data;

[0113] S14, safety monitoring data collection: smoke composition sensors and fire alarms are configured in emission monitoring facilities and fire safety units to collect safety monitoring data;

[0114] S15, production process data collection: fuel consumption sensors and water quality monitoring sensors are set in the fuel supply and water treatment units to collect production process data;

[0115] S16, real-time operation data collection: deploy relevant sensors and data interfaces in the power plant management unit to collect production plan execution and equipment maintenance status in real time;

[0116] Through the above content, comprehensive real-time monitoring of electrical parameters, environmental status, equipment health, safety monitoring, production processes and operation data is achieved, ensuring the comprehensiveness, accuracy and real-time nature of data collection, and supporting the efficient operation and safety management of power plants.

[0117] Data preprocessing in S2 includes:

[0118] S21, data denoising: The wavelet transform algorithm is used to perform multi-scale analysis on the digital power plant data, decompose the data into different frequency bands, and perform threshold processing on the high-frequency part to remove noise, which is expressed as:

[0119] ;

[0120] in, is the reconstructed signal, is the wavelet transform coefficient, is the wavelet basis function, and are the number of decomposition layers and time steps respectively;

[0121] S22, missing value filling: For the missing values ​​in the digital power plant data, the K nearest neighbor algorithm is used to fill the missing values. The nearest data point is selected by calculating the Euclidean distance between the data point and its surrounding known data points. The weighted average of the data points is expressed as:

[0122] ;

[0123] in, For the filled data points, For the most recent known data points, is the Euclidean distance between the corresponding data point and the missing point;

[0124] Unified data format: Regular expression matching and conversion algorithms are used to standardize data formats into unified timestamps, units, and data types to ensure compatibility and consistency of data during storage and analysis;

[0125] Through the above content, not only the efficiency and accuracy of subsequent data storage and analysis are improved, but also a solid data foundation is provided for the intelligent management and decision support of digital power plants, ensuring the stability and reliability of the system in complex environments.

[0126] Data relationship modeling in S31 includes:

[0127] S311, Graph Structure Construction: Use Graph Neural Network (GNN) to model the time series and spatial position relationship of digital power plant data from different sensors, and regard each sensor data as a node in the graph , the edges between nodes Represents the association between sensor data and the weight of the edge The calculation is based on the similarity of the time series and the proximity of the spatial position, expressed as:

[0128] ;

[0129] in, For sensor and The similarity in time series, For sensor and The distance in space, and To adjust the hyperparameters of weights;

[0130] S312, node feature initialization: each node The initial eigenvector of It is obtained by extracting the features of the corresponding sensor data, including the statistical features of the sensor measurement values ​​and the sensor location and type, and is expressed as:

[0131] ;

[0132] in, and Respectively represent sensors The mean and standard deviation of the collected data, Indicates sensor spatial location, Indicates the type of sensor;

[0133] S313, graph convolution operation: feature propagation and update of nodes in the graph structure through the graph convolutional neural network (GCN), expressed as:

[0134] ;

[0135] in, Indicates Node in layer The characteristic vector of For Node The set of neighbor nodes of For the The weight matrix of the layer, is the ReLU activation function, and For Node and The degree of

[0136] S314, Potential Association Identification: Through layer-by-layer feature updates of multi-layer graph convolutional neural networks (GCN), a high-dimensional feature representation of each node is finally obtained. , combined with the weights of the edges in the graph Identify potential correlations between sensor data, expressed as:

[0137] ;

[0138] in, is the final incidence matrix, is the sigmoid activation function, which is used to map the association strength to the [0,1] interval. and They are respectively Post-layer node and nodes The final eigenvector of ;

[0139] Through the above content, it is possible to capture complex spatiotemporal relationships, provide accurate data association analysis, and improve the intelligence level of data storage and the accuracy of retrieval. Through the multi-layer feature extraction of graph neural networks, the system can dynamically adapt to changes in the operating status of the power plant, ensuring consistency and correlation in the data storage and analysis process, thereby supporting more accurate operation optimization and fault diagnosis.

[0140] Dynamic association optimizations in S32 include:

[0141] S321, Real-time status monitoring and input update: Real-time monitoring of the operating status parameters of the power plant, including load changes, environmental conditions, and equipment status, and using the operating status parameters as dynamic inputs to the graph neural network (GNN) to update the feature representation of each node , expressed as:

[0142] ;

[0143] in, For Node In the The feature vector after layer graph convolution, is the ReLU activation function, For Node The set of neighbor nodes of For Node and nodes The edge weights between is the mapping function, according to the current power plant operation status Adjust edge weights , and Node and nodes The degree of For the The weight matrix of the layer, Neighbor Node In the Feature vector after layer graph convolution;

[0144] S322, dynamic adjustment of edge weights: adjust the node weights according to the real-time changes in the power plant operation status. and nodes The edge weight between , so that the association relationship between nodes dynamically reflects the current operating conditions of the power plant. For example, when the load increases, the edge weight of the relevant sensor data will be increased to highlight its importance in data storage and analysis, which is expressed as:

[0145] ;

[0146] in, is the initial edge weight;

[0147] S323, Multi-layer Graph Convolutional Network Optimization: After iterative updates of multi-layer graph convolutional networks (GCN), node feature representation The edge weights between nodes Dynamically reflects the optimal correlation relationship under the current power plant operation status, and outputs node features and edge weights to construct an optimized correlation matrix , expressed as:

[0148] ;

[0149] in, is the sigmoid activation function, For Node In the The final feature vector after layer graph convolution, For Node In the The final feature vector after layer graph convolution;

[0150] Through the above content, it is possible to flexibly respond to changes in power plant load, environmental conditions, and equipment health status, ensuring that in a complex and changeable operating environment, data associations always accurately reflect the current system status. By dynamically adjusting edge weights, the accuracy of data storage and analysis is improved, supporting more intelligent fault warnings, operation optimization, and resource allocation, ultimately enhancing the overall operating efficiency and reliability of digital power plants.

[0151] Mapping Function It is expressed as:

[0152] ;

[0153] in, They are the weight factors of load, environmental conditions, and equipment status, reflecting the relative importance of each indicator to the power plant status. is the current load level, divided by the maximum load Normalization is performed so that the result is interval, is the current environmental condition index, divided by the reference environmental value Perform normalization processing, Current device health divided by maximum health Perform normalization.

[0154] The associated data stores in S33 include:

[0155] S331, Data bundling strategy generation: Based on the association matrix optimized by graph neural network (GNN) , determine the set of data nodes with strong associations , if the optimized correlation matrix The value exceeds the preset threshold , then the corresponding node and nodes Join the same collection ;

[0156] Preset Threshold Based on the association distribution setting, it specifically includes:

[0157] Calculate the correlation: The correlation matrix calculated by the graph neural network , determine the association degree between all node pairs;

[0158] Construct the correlation distribution: All calculated correlations are Sort them in ascending order and draw a correlation distribution graph to observe the distribution characteristics of the correlation;

[0159] Setting Thresholds :According to the correlation distribution graph, select an appropriate threshold , expressed as:

[0160] ;

[0161] in, Indicates that the correlation exceeds the threshold The proportion of node pairs, is a preset ratio value (such as );

[0162] S332, data block division and bundled storage: each associated data set The digital power plant data after correlation optimization is packaged into independent data blocks , and separate data blocks To perform bundle storage;

[0163] Through the above content, the digital power plant data that has been optimized for association is bundled and stored, which not only maintains the logical association of the data, but also optimizes the data access efficiency, ensuring that relevant data can be accessed quickly and accurately in complex query and analysis tasks.

[0164] Data compression and encryption in S4 include:

[0165] S41, data compression: A hybrid compression algorithm is used to compress the associated stored digital power plant data, Huffman coding is used to compress the real-time digital power plant data, and discrete wavelet transform is used to compress the historical digital power plant data, which is expressed as:

[0166] ;

[0167] in, is the average code length after Huffman coding, is the total number of symbols, For symbol The probability of appearing in the data, For symbol The corresponding codeword length in Huffman coding;

[0168] ;

[0169] in, is the compressed data after discrete wavelet transform and quantization. is the number of wavelet decomposition layers, is the approximate coefficient after wavelet decomposition in The value of the layer, is the detail coefficient after wavelet decomposition in The value of the layer, Represents the detail coefficient The quantitative processing performed;

[0170] S42, Data Encryption: The compressed digital power plant data is encrypted using the AES algorithm. The encryption process includes generating a symmetric key and Block encryption is performed to ensure that data is protected from unauthorized access and tampering during storage and transmission, as shown below:

[0171] ;

[0172] in, is the encrypted data block, is the compressed data block, is the symmetric encryption key, is the encryption function;

[0173] Through the above content, the storage space occupied by digital power plant data can be effectively reduced, while ensuring the security of data during transmission and storage, providing reliable protection for the subsequent analysis and application of data.

[0174] The adaptive redundancy strategy optimization algorithms in S5 include:

[0175] S51, data importance weight calculation: for each type of digital power plant data , calculate its importance weight in power plant operation , expressed as:

[0176] ;

[0177] in, To meet the real-time data requirements, For the criticality of data, and is the weight coefficient, which indicates the influence of real-time and criticality on the importance of data;

[0178] S52, data access frequency calculation: by recording the data access log, calculate each data block Frequency of visits , represents the number of times the data block is accessed within the time window;

[0179] S53, Dynamic adjustment of redundancy level: based on the importance weight of the data and access frequency , dynamically adjust each data block Redundancy level , ensuring that important and frequently accessed data has a higher level of redundancy, expressed as:

[0180] ;

[0181] in, For data blocks The number of redundant copies of is the adjustment factor used to control the overall level of global redundancy level, is the number of all data blocks;

[0182] S54, Dynamic Allocation and Storage: Based on Calculated Redundancy Level , for data blocks Dynamically allocate a corresponding number of redundant copies and distribute them to different storage nodes to improve data reliability and availability;

[0183] Through the above content, the redundancy level can be dynamically adjusted according to the importance and access frequency of the digital power plant data, ensuring that critical data has sufficient redundancy protection during fault recovery and high-frequency access, while optimizing the utilization efficiency of storage resources.

[0184] Data retrieval and analysis in S6 include:

[0185] S61, Establishment of an indexing mechanism based on a graph structure: By analyzing the graph structure of the digital power plant data, a multi-level indexing mechanism is constructed, which can quickly locate the target data node and its associated context data, including:

[0186] S611, global index: partition the entire graph structure, each partition contains an associated set of nodes, which is used to locate the partition where the target node is located;

[0187] S612, local index: in each partition, based on the edge weights between nodes, establish a local index for retrieving context data associated with the target node;

[0188] S62, fast data retrieval algorithm: using the index mechanism based on the graph structure, locate the partition where the target data node is located through the global index, and retrieve the context data closely related to the target node through the local index, expressed as:

[0189] ;

[0190] in, Representation Node The context data set of the node The correlation exceeds the threshold All nodes A collection of Representation Node and nodes The edge weights between is a preset edge weight threshold used to filter out nodes with low correlation to ensure the relevance and accuracy of the retrieval results. Representation Node The set of neighbor nodes of

[0191] S63, context data analysis: Analyze the retrieved context data, use graph neural network (GNN) to mine the complex correlation between data, and combine the operation status of the power plant to optimize the operation and diagnose faults of the power plant, including:

[0192] S631, feature extraction: extract features from context data, including time series features, equipment operation status features, data association features, failure mode features, spatial and topological features, and access frequency features, and generate context feature vectors , expressed as:

[0193] ;

[0194] in, For Node and nodes The edge weights between The context node The eigenvector of

[0195] S632, relational reasoning: Use graph neural network to reason about context feature vectors and derive the correlation and fault cause with target data nodes, expressed as:

[0196] ;

[0197] in, For Node and nodes The inference relationship score between and Respectively represent nodes and nodes go through The final feature vector after layer graph convolution, is the weight matrix, is the bias term, is the sigmoid activation function;

[0198] Through the above content, the speed and accuracy of data retrieval are improved. At the same time, through comprehensive analysis of contextual data, potential failure risks and optimization opportunities can be more comprehensively identified, supporting the intelligent operation and preventive maintenance of power plants, and effectively improving the operating efficiency and safety of power plants.

[0199] like Figure 2 As shown, a digital power plant data automatic storage system is used to implement the above-mentioned digital power plant data automatic storage method, including the following modules:

[0200] Data acquisition module: uses multiple sensors deployed in the digital power plant to collect digital power plant data in real time, including electrical parameter data, environmental parameter data, equipment status data, safety monitoring data, production process data and real-time operation data;

[0201] Data preprocessing module: preprocess the collected digital power plant data, including data denoising, missing value filling and data format unification;

[0202] Context-associated storage module: Use graph neural network to perform context-associated analysis on pre-processed digital power plant data and perform associative storage;

[0203] Data compression and encryption module: compress and encrypt the associated stored digital power plant data;

[0204] Data synchronization and backup module: regularly synchronize and back up the digital power plant data stored in the database. Adopt an adaptive redundancy strategy optimization algorithm during the synchronization and backup process to dynamically adjust the redundancy level according to the importance and access frequency of the data.

[0205] Data retrieval and analysis module: By establishing a graph-based indexing mechanism, contextual data related to the target digital power plant data can be quickly retrieved and analyzed to support power plant operation optimization and fault diagnosis.

[0206] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0207] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for automatically storing digital power plant data, characterized in that: The following steps are involved: S1, data collection: using multiple sensors deployed in the digital power plant to collect digital power plant data in real time, including electrical parameter data, environmental parameter data, equipment status data, safety monitoring data, production process data and real-time operation data; S2, data preprocessing: preprocessing the collected digital power plant data, including data denoising, missing value filling and data format unification; S3, context-related storage: Context-related analysis of pre-processed digital power plant data is performed through graph neural networks, including: S31, data relationship modeling: Use graph neural network to model the time series and spatial position relationship of digital power plant data from different sensors, build a graph structure between data, and identify potential associations between different data sources. The data relationship modeling includes: S311, graph structure construction: Use graph neural network to model the time series and spatial position relationship of digital power plant data from different sensors, and regard each sensor data as a node in the graph , the edges between nodes Represents the association between sensor data and the weight of the edge The calculation is based on the similarity of the time series and the proximity of the spatial position, expressed as: ； in, For sensor and The similarity in time series, For sensor and The distance in space, and To adjust the hyperparameters of weights; S312, node feature initialization: each node The initial eigenvector of It is obtained by extracting the features of the corresponding sensor data, including the statistical features of the sensor measurement values ​​and the sensor location and type, expressed as: ； in, Respectively represent sensors The mean and standard deviation of the collected data, Indicates sensor spatial location, Indicates the type of sensor; S313, graph convolution operation: feature propagation and update of nodes in the graph structure through the graph convolutional neural network, expressed as: ； in, Indicates Node in layer The characteristic vector of For Node The set of neighbor nodes of is the weight matrix of the kth layer, is the ReLU activation function, and For Node and The degree of S314, Potential Association Identification: Through layer-by-layer feature updates of multi-layer graph convolutional neural networks, high-dimensional feature representation of each node is finally obtained. , combined with the weights of the edges in the graph Identify potential correlations between sensor data, expressed as: ； in, is the final incidence matrix, is the sigmoid activation function, and They are the nodes after the Lth layer of the graph neural network. and nodes The final eigenvector of ; S32, dynamic association optimization: according to the real-time changes in the operating status of the power plant, the association relationship between the digital power plant data is dynamically adjusted through the graph neural network. The dynamic association optimization includes: S321, Real-time status monitoring and input update: Real-time monitoring of the operating status parameters of the power plant, including load changes, environmental conditions, and equipment status, and using the operating status parameters as dynamic inputs to the graph neural network to update the feature representation of each node , expressed as: ； in, For Node In the The feature vector after layer graph convolution, is the ReLU activation function, For Node The set of neighbor nodes of For Node and nodes The edge weights between is the mapping function, according to the current power plant operation status Adjust edge weights , and Node and nodes The degree of is the weight matrix of the kth layer, Neighbor Node Feature vector after the k-th layer of graph convolution; S322, dynamic adjustment of edge weights: adjust the node weights according to the real-time changes in the power plant operation status. and nodes The edge weight between , so that the association relationship between nodes dynamically reflects the current operating conditions of the power plant, expressed as: ； in, is the initial edge weight; S323, Multi-layer graph convolutional network optimization: After iterative updates of multi-layer graph convolutional networks, node feature representation The edge weights between nodes Dynamically reflects the optimal correlation relationship under the current power plant operation status, and outputs node features and edge weights to construct an optimized correlation matrix , expressed as: ； in, is the sigmoid activation function, For Node The final feature vector after the Lth layer of graph convolution, For Node The final feature vector after the Lth layer of graph convolution; S33, associated data storage: bundling and storing the associated optimized digital power plant data; S4, data compression and encryption: compressing and encrypting the associated stored digital power plant data; S5, data synchronization and backup: regularly synchronize and back up the digital power plant data stored in the database. During the synchronization and backup process, an adaptive redundancy strategy optimization algorithm is used to dynamically adjust the redundancy level according to the importance and access frequency of the data. The adaptive redundancy strategy optimization algorithm includes: S51, data importance weight calculation: for each type of digital power plant data , calculate its importance weight in power plant operation , expressed as: ； in, To meet the real-time data requirements, Because of the criticality of the data, and is the weight coefficient; S52, data access frequency calculation: by recording the data access log, calculate each data block Frequency of visits ; S53, Dynamic adjustment of redundancy level: based on the importance weight of the data and access frequency , dynamically adjust each data block Redundancy level , expressed as: ； in, For data blocks The number of redundant copies of is the adjustment factor, N is the number of all data blocks; S54, Dynamic Allocation and Storage: Based on Calculated Redundancy Level , for data blocks Dynamically allocate a corresponding number of redundant copies and distribute the redundant copies to different storage nodes; S6, data retrieval and analysis: By establishing a graph-based indexing mechanism, contextual data related to the target digital power plant data can be quickly retrieved and analyzed to support power plant operation optimization and fault diagnosis.

2. The automatic storage method of digital power plant data according to claim 1, characterized in that: The data collection in S1 includes: S11, electrical parameter data collection: deploy current sensors, voltage sensors and frequency sensors in power generation equipment and distribution facilities to collect electrical parameter data in real time; S12, Environmental parameter data collection: Deploy temperature sensors, pressure sensors, and humidity sensors in boilers, pipelines, and cooling facilities to monitor environmental parameters; S13, equipment status data collection: installing vibration sensors, speed sensors, oil level sensors and oil temperature sensors on mechanical equipment to obtain equipment status data; S14, safety monitoring data collection: smoke composition sensors and fire alarms are configured in emission monitoring facilities and fire safety units to collect safety monitoring data; S15, production process data collection: fuel consumption sensors and water quality monitoring sensors are set in the fuel supply and water treatment units to collect production process data; S16, real-time operation data collection: Deploy relevant sensors and data interfaces in the power plant management unit to collect production plan execution and equipment maintenance status in real time.

3. The automatic storage method of digital power plant data according to claim 1, characterized in that: The data preprocessing in S2 includes: S21, data denoising: wavelet transform algorithm is used to perform multi-scale analysis on digital power plant data, decompose the data into different frequency bands, and perform threshold processing on the high-frequency part to remove noise; S22, missing value filling: For the missing values ​​in the digital power plant data, the K nearest neighbor algorithm is used to fill them. By calculating the Euclidean distance between the data point and the surrounding known data points, the nearest K data points are selected for weighted averaging; Data format unification: Regular expression matching and conversion algorithms are used to standardize the data format into a unified timestamp, unit, and data type.

4. The automatic storage method of digital power plant data according to claim 1 is characterized in that: The associated data storage in S33 includes: S331, Data bundling strategy generation: Based on the association matrix optimized by graph neural network , determine the set of data nodes with strong associations , if the optimized correlation matrix The value exceeds the preset threshold , then the corresponding node and nodes Join the same collection ; S332, data block division and bundled storage: each associated data set The digital power plant data after correlation optimization is packaged into independent data blocks , and separate data blocks Bundle storage.

5. The automatic storage method of digital power plant data according to claim 1, characterized in that: The data compression and encryption in S4 include: S41, data compression: use a hybrid compression algorithm to compress the associated stored digital power plant data, use Huffman coding to compress the real-time digital power plant data, and use discrete wavelet transform to compress the historical digital power plant data; S42, data encryption: The compressed digital power plant data is encrypted using a standard encryption algorithm. The encryption process includes generating a symmetric key and Block encryption is performed, expressed as: ； in, is the encrypted data block, is the compressed data block, K is the symmetric encryption key, is the encryption function.

6. The automatic storage method of digital power plant data according to claim 1, characterized in that: The data retrieval and analysis in S6 includes: S61, Establishment of an indexing mechanism based on a graph structure: By analyzing the graph structure of digital power plant data, a multi-level indexing mechanism is constructed, including: S611, global index: partition the entire graph structure, each partition contains an associated set of nodes, which is used to locate the partition where the target node is located; S612, local index: in each partition, based on the edge weights between nodes, establish a local index for retrieving context data associated with the target node; S62, fast data retrieval algorithm: using the index mechanism based on the graph structure, locate the partition where the target data node is located through the global index, and retrieve the context data closely related to the target node through the local index, expressed as: ； in, Representation Node A collection of context data, Representation Node and nodes The edge weights between is the preset edge weight threshold, Representation Node The set of neighbor nodes of S63, context data analysis: Analyze the retrieved context data, use graph neural network to mine the complex correlation between data, and combine the operation status of the power plant to optimize the operation and diagnose faults of the power plant, including: S631, feature extraction: extract features from context data, including time series features, equipment operation status features, data association features, failure mode features, spatial and topological features, and access frequency features, and generate context feature vectors , expressed as: ； in, For Node and nodes The edge weights between The context node The eigenvector of S632, relational reasoning: Use graph neural network to reason about context feature vectors and derive the correlation and fault cause with target data nodes, expressed as: ； in, For Node and nodes The inference relationship score between and Respectively represent nodes and nodes The final feature vector after L layers of graph convolution, W is the weight matrix, b is the bias term, is the sigmoid activation function.

7. An automatic storage system for digital power plant data, used to implement an automatic storage method for digital power plant data as claimed in any one of claims 1 to 6, characterized in that: Includes the following modules: Data acquisition module: uses multiple sensors deployed in the digital power plant to collect digital power plant data in real time, including electrical parameter data, environmental parameter data, equipment status data, safety monitoring data, production process data and real-time operation data; Data preprocessing module: preprocess the collected digital power plant data, including data denoising, missing value filling and data format unification; Context-associated storage module: Use graph neural network to perform context-associated analysis on pre-processed digital power plant data and perform associative storage; Data compression and encryption module: compress and encrypt the associated stored digital power plant data; Data synchronization and backup module: regularly synchronize and back up the digital power plant data stored in the database. Adopt an adaptive redundancy strategy optimization algorithm during the synchronization and backup process to dynamically adjust the redundancy level according to the importance and access frequency of the data. Data retrieval and analysis module: By establishing a graph-based indexing mechanism, contextual data related to the target digital power plant data can be quickly retrieved and analyzed to support power plant operation optimization and fault diagnosis.

Images