New energy power station intelligent auxiliary control system based on digital twinning

Abstract

The invention discloses a new energy power station intelligent auxiliary control system based on digital twinning, and the system comprises a data collection layer which collects physical entity data, including equipment operation data, environment data and power grid data, of a new energy power station, the environment data is collected through environment monitoring equipment, and the power grid data is collected through a power grid side sensor; the data processing layer is used for carrying out preprocessing, feature extraction and data fusion on the collected data, the preprocessing comprises denoising, missing value filling and abnormal value processing, and the feature extraction comprises statistical feature extraction, frequency domain feature extraction and feature extraction based on a physical model; the method has the beneficial effects that real-time monitoring and predictive maintenance are realized through a digital twin technology, the fault response time is shortened, the non-planned shutdown frequency is reduced, and the operation and maintenance cost is reduced; through multi-source data fusion and deep learning model application, the generating capacity prediction error is reduced, and scheduling margin redundancy is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent management of new energy power stations, and specifically relates to an intelligent auxiliary control system for new energy power stations based on digital twins. Background Art

[0002] With the growth of global energy demand and the intensification of environmental problems, new energy, as a clean, renewable, low-carbon energy source, has received widespread attention and attention from governments and societies around the world. New energy power stations refer to power generation facilities built using new energy resources such as wind energy, solar energy, and biomass energy. They have the advantages of saving energy, reducing pollution, and protecting the ecological environment. They are an important way to achieve energy transformation and sustainable development.

[0003] However, new energy power stations also face many challenges. On the one hand, the output of new energy power stations is intermittent, volatile, and uncertain, which brings tremendous pressure to the dispatching, operation, and control of the power system; on the other hand, the scale of new energy power stations continues to expand, the types of equipment are diverse, and the operation and maintenance tasks are complex.

[0004] Patent publication number CN118983951B is a comprehensive auxiliary control system based on multiple parameters of a smart substation. The patent discloses a load range determination module, an energy loss test module, a tap flexibility analysis module, a multi-parameter balance auxiliary control module and an optimization module; the load range determination module is used to monitor the load-end voltage demand required by the corresponding transformer in the substation in advance using a monitoring instrument, and obtain the range of the corresponding transformer output voltage based on the load distribution state analysis; the energy loss test module, based on the range of the corresponding transformer output voltage, sets different tap number conditions on the winding of the corresponding transformer for multiple tests, and during the test process, monitors in real time the switching energy loss Nsz, heat loss Rsz and transmission loss Csz caused by the instantaneous changes in voltage and current when switching taps under different tap number conditions, and analyzes and calculates the stability factor Wdyz based on the switching energy loss Nsz and heat loss Rsz. If the stability factor Wdyz exceeds the preset threshold, a tap number screening instruction is issued.

[0005] The traditional operation and maintenance model of new energy power stations has many problems: low operation and maintenance efficiency, reliance on manual inspections, and a high rate of missed detection of hidden faults; insufficient forecasting capabilities and large errors in power generation forecasts; serious data silos and low interoperability of multi-system data, making it difficult to achieve efficient data analysis and decision support. These problems have seriously restricted the intelligent development and efficient operation of new energy power stations. Summary of the Invention

[0006] The purpose of the present invention is to provide an intelligent auxiliary control system for new energy power stations based on digital twins to solve the problems of low operation and maintenance efficiency, insufficient prediction capabilities, serious data islands, etc. in the existing technology of new energy power stations, realize real-time monitoring, intelligent analysis, remote control and optimized scheduling of new energy power stations, and improve the operating efficiency, reliability and economy of new energy power stations.

[0007] To achieve the above objectives, the present invention provides the following technical solutions: an intelligent auxiliary control system for a new energy power station based on digital twins, comprising The data collection layer collects physical entity data of the new energy power station, including equipment operation data, environmental data, and power grid data. The equipment operation data is collected by sensors deployed on the equipment, the environmental data is collected by environmental monitoring equipment, and the power grid data is collected by grid-side sensors. The data processing layer performs preprocessing, feature extraction, and data fusion on the collected data. The preprocessing includes denoising, missing value filling, and outlier processing. The feature extraction includes statistical features, frequency domain features, and physical model-based feature extraction. The data fusion standardizes multi-source heterogeneous data and organizes and stores them by subject and dimension. The digital twin model layer builds a digital twin model of the new energy power station, including a real-world 3D model of the station and an equipment information model. The real-world 3D model of the station is constructed using data collected by drone aerial photography and ground laser scanning technology, and the equipment information model integrates static and dynamic information of the equipment. The intelligent analysis layer implements operating status analysis, fault diagnosis, power generation forecasting, and optimized scheduling based on the digital twin model and data analysis algorithms. The data analysis algorithms include data mining algorithms and machine learning algorithms. The machine learning algorithms are used to build power generation forecasting models and fault diagnosis models. The application service layer provides users with a visual interface and interaction, including 3D visualization scenes and data dashboards, enabling remote monitoring, fault warning, maintenance decision-making, and optimized scheduling, and supports multi-terminal access. The edge computing layer, deployed in edge monitoring units, completes communication access, edge analysis, and data aggregation. Edge gateways are deployed on wind turbine towers and photovoltaic combiner boxes, supporting MQTT and CoAP protocols to collect sensor data in real time and perform local preprocessing. The data backplane layer builds the system's data entity-relationship model, including power plant basic information entities, equipment asset information entities, equipment monitoring data entities, equipment fault record entities, equipment maintenance record entities, and system user information entities. It uses a relational database to implement structured storage and associative query of data. The model optimization module optimizes the digital twin model, including mesh simplification, physical field coupling, and model rendering optimization. It reduces the number of model faces through mesh simplification, achieves real-time physical mapping of the device's operating status through physical field coupling, and enables lightweight loading and cross-platform display of the model through WebGL technology. The predictive maintenance module generates a predictive maintenance plan based on an equipment health assessment model and a remaining life prediction model. The equipment health assessment model integrates multi-parameter data to generate an equipment health index. The remaining life prediction model predicts the remaining life of the equipment based on the Weibull distribution and optimizes spare parts management and maintenance strategies.

[0008] As a preferred technical solution of the present invention, it also includes an information infrastructure layer, which is responsible for data collection, transmission, storage and calculation, including a power plant control system, computing resources, network facilities and a data perception network. The power plant control system is responsible for the daily operation and management of the power plant. The computing resources provide high-performance computing capabilities and support complex data analysis and model simulation. The network facilities ensure real-time transmission of data. The data perception network covers power generation, transmission, distribution, transformation and energy storage links, forming a complete data collection system.

[0009] As a preferred technical solution of the present invention, it also includes a digital twin technology layer, which realizes intelligent management of power stations in a data-driven manner, and includes three major modules: model and algorithm, visualization model, and data baseboard; the model and algorithm module uses data analysis, power generation prediction and fault prediction to monitor the operating status of the power station in real time and predict trends; the visualization model module converts complex power station operation data into an intuitive graphical interface through three-dimensional modeling and multi-dimensional data display; the data baseboard module is responsible for data aggregation, governance and mining, and provides stable data support for the upper-level model.

[0010] As a preferred technical solution of the present invention, it also includes a user operation layer, an interface for direct interaction between the system and the user, providing a graphical user interface for operators to perform monitoring, control and management tasks, adopting modern Web technology, supporting access to the system from a variety of terminal devices, and can also display the digital twin model of the power station in a graphical manner.

[0011] As a preferred technical solution of the present invention, it also includes a data storage layer, which is responsible for the storage and management of all data in the system, including real-time monitoring data, historical data, system configuration and user information, and adopts a relational database management system or a NoSQL database to achieve efficient storage, fast retrieval and secure access to data, and is responsible for providing data support for the digital twin model, including real-time data, historical data, model parameters and simulation results.

[0012] As a preferred technical solution of the present invention, it also includes a business logic layer, which processes requests from the user operation layer and performs specific business logic processing. It is responsible for interacting with the data storage layer to obtain or update the stored data. By using digital twin technology, by connecting the physical entity with its digital representation, a virtual model of the new energy power station is constructed using physical models, sensor data and historical data.

[0013] As a preferred technical solution of the present invention, the process of constructing the real-scene three-dimensional model of the station includes five stages: data collection, processing, modeling, optimization and verification.

[0014] As a preferred technical solution of the present invention, it also includes statistical analysis of new energy power stations, which is achieved through statistical analysis models and algorithms.

[0015] Compared with the prior art, the present invention has the following beneficial effects: Real-time monitoring and predictive maintenance are achieved through digital twin technology, shortening fault response time, reducing unplanned downtime, and lowering operation and maintenance costs. Multi-source data fusion and deep learning model application can reduce power generation forecast errors and dispatch margin redundancy; The unified data platform improves the interoperability of multi-system data, shortens cross-system analysis time, and improves decision-making efficiency; Through intelligent scheduling and equipment optimization control, the wind power pneumatic efficiency is improved, the photovoltaic module power generation efficiency is improved, and the wind curtailment rate is reduced; The perfect security protection mechanism ensures data security and stable system operation, and improves the reliability and safety of new energy power stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is the overall framework diagram of the visualization platform of the present invention; Figure 2 This is a schematic diagram of the user operation layer design of the present invention; Figure 3 This is a schematic diagram of the business logic layer structure of the present invention. DETAILED DESCRIPTION

[0017] 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.

[0018] See also Figure 1-Figure 3A digital twin-based intelligent auxiliary control system for new energy power plants adopts a multi-layer architecture approach based on digital twin technology to cope with the complexity and dynamic nature of new energy power plant management. The system includes an overall architecture, a user operation layer, a business logic layer, and a data storage layer, aiming to improve the operation and maintenance efficiency and decision-making quality of new energy power plants. Overall architecture design: The design of the intelligent auxiliary control system for new energy power plants utilizes a multi-layered architecture model based on digital twin technology, aiming to provide efficient and reliable intelligent monitoring and management solutions for new energy power plants. The overall design is based on physical entities and, supported by information infrastructure, builds a complete digital twin technology system to achieve efficient management and intelligent operation and maintenance of new energy power plants. The core of this architecture is to map the operating status of the power plant in the physical world to the digital space in real time, and optimize operational decisions through a data-driven approach to improve the reliability and cost-effectiveness of the power plant. The physical site layer is the foundation of the entire system, encompassing wind farms, photovoltaic power plants, and other types of power generation facilities. The operating status, equipment parameters, and environmental data of these physical power plants form the data source for the digital twin. Through sensor networks and monitoring equipment, real-time operational data from the power plants is collected and transmitted to the information infrastructure layer, providing a foundation for subsequent data analysis and model building. The design of this layer ensures a seamless connection between the physical and digital worlds, enabling the digital twin model to accurately reflect the actual status of the power plant. The information infrastructure layer is the key support for the digital twin system, responsible for data collection, transmission, storage, and calculation. This layer includes the power plant control system, computing resources, network facilities, and data perception network. The power plant control system is responsible for the daily operation and management of the power plant, ensuring the stability and safety of the power generation process. The computing resources provide high-performance computing power to support complex data analysis and model simulation. The network facilities ensure real-time data transmission and ensure the feasibility of remote monitoring and operation and maintenance. The data perception network covers power generation, transmission, distribution, transformation, and energy storage, forming a complete data collection system and providing comprehensive data support for the digital twin model. The design of this layer ensures the integrity, real-timeness, and security of the data, laying a solid foundation for the application of digital twin technology. The digital twin technology layer is the core of the entire architecture, enabling intelligent management of power plants through a data-driven approach. This layer comprises three modules: models and algorithms, visualization models, and a data backplane. The model and algorithm module utilizes data analysis, power generation forecasting, and fault prediction technologies to monitor the power plant's operating status in real time and predict trends, providing a scientific basis for operation and maintenance decisions. The visualization model module transforms complex power plant operating data into an intuitive graphical interface through three-dimensional modeling and multi-dimensional data display, enabling managers to quickly understand the power plant's status. The data backplane module is responsible for data aggregation, governance, and mining, ensuring data quality and availability while providing stable data support for upper-layer models. The design of this layer enables the digital twin system to not only reflect the power plant's operating status in real time, but also improve its overall efficiency through predictive maintenance and optimized control. The entire architecture is designed based on the principles of modularity and scalability, ensuring that the system can adapt to different types of new energy power plants and flexibly adjust as business needs change. Through a layered design, the system achieves a deep integration of the physical and digital worlds, shifting power plant management from traditional passive operation and maintenance to active prediction and optimization control. The application of digital twin technology not only improves the operating efficiency of power plants, but also reduces operation and maintenance costs, providing strong technical support for the intelligent development of new energy power plants.

[0019] Visualization platform design: The design of the intelligent auxiliary control system for new energy power plants adopts a multi-layered architecture model based on digital twin technology, aiming to provide efficient and reliable intelligent monitoring and management solutions for new energy power plants. This architecture is divided into a user operation layer, a business logic layer, and a data storage layer. Each layer has independent responsibilities and functions, while working together to ensure the overall performance and stability of the system. The user operation layer (Presentation Layer) is the interface for direct interaction between the system and the user, providing a graphical user interface (GUI) for operators to perform monitoring, control, and management tasks. By adopting modern web technologies, a responsive, intuitive, and user-friendly operating environment is implemented, supporting access to the system from a variety of terminal devices (including PCs, tablets, and smartphones), allowing operators to monitor and manage power plant status in real time regardless of their location. The user operation layer can also graphically display the power plant's digital twin model, allowing users to intuitively understand the power plant's real-time status, historical trends, and future forecasts, as well as remotely control and optimize the plant. Business Logic Layer: As the core of the system, this layer handles requests from the user operation layer and performs specific business logic processing, such as data analysis, fault diagnosis, and optimization suggestions. This layer is also responsible for interacting with the data storage layer to obtain or update stored data. By adopting a modular design, the business logic layer can flexibly respond to different business needs while facilitating future functional expansion and maintenance. The core function of the business logic layer is to use digital twin technology to connect physical entities with their digital representations, and use physical models, sensor data, and historical data to build a virtual model of a new energy power station. This can not only achieve real-time mapping and simulation of power station properties, status, and behavior, but also predict future operating trends and provide decision support for operations management. The Data Storage Layer is responsible for the storage and management of all data in the system, including real-time monitoring data, historical data, system configuration, and user information. A relational database management system or NoSQL database is used to achieve efficient data storage, fast retrieval, and secure access. The design of the data storage layer takes into account data consistency, integrity, and backup and recovery mechanisms, ensuring the reliability and security of system data. The data storage layer is also responsible for providing data support for the digital twin model, including real-time data, historical data, model parameters, simulation results, etc., to ensure the accuracy and real-time performance of the digital twin model. The design of the entire system is based on the principles of service-oriented architecture (SOA) and microservice architecture, supporting a high degree of modularity and scalability, facilitating system maintenance and upgrades; in addition, the system architecture adopts a variety of security measures, including data encryption, user authentication and authorization management, to ensure the security of the system and data.

[0020] User operation layer design: In the intelligent auxiliary control system of new energy power plants, the design of the user operation layer is crucial, as it directly affects the user's overall system efficiency and experience. This layer utilizes digital twin technology and adopts mainstream front-end development technologies such as JSP, HTML, and CSS to ensure that the system interface maintains excellent compatibility and responsiveness across various devices and browsers. The design adheres to the four principles of intuitiveness, consistency, responsiveness, and efficiency, aiming to create an aesthetically pleasing, practical, easy-to-use, and efficient user operation environment. The principle of intuitiveness: through the visual interface of the digital twin, even users who are using the system for the first time can quickly understand how to operate; the system provides intuitive virtual model operations, and tasks can be completed without reading complex help documents; unified interface style, color matching and layout typesetting reduce the user's learning cost and enhance the overall coordination of the system; the consistent display of the digital twin model ensures that users can intuitively see the operating status of the power station and any changes; the system ensures that every user operation receives an immediate and clear response through real-time data feedback; whether it is confirmation of successful operation or error prompt, the digital twin model can clearly let users know the current system status and any potential problems; by simplifying the operation process, optimizing the page layout and using digital twin technology to automatically adjust the control strategy, the speed at which users complete tasks is greatly improved; asynchronous data loading technology is applied to the dynamic update of data, significantly improving the page response speed and the smoothness of user operation; To further optimize performance, caching strategies and dynamic content update mechanisms were introduced to reduce server burden and accelerate user access. System security was enhanced through multiple security technologies such as data encryption, user authentication, and access control, ensuring the security of user data and the reliability of system operations. In terms of the functional layout of the user operation layer, whether it is real-time video monitoring, security management, environmental monitoring, visual management or system settings, the operation process is simplified through clear layout and intuitive navigation elements. At the same time, the responsiveness of control elements such as buttons and forms is optimized, greatly improving the user experience. To quantitatively evaluate the performance of the user operation layer, a performance scoring formula, UOP, was introduced. This formula comprehensively considers the weight of various indicators and the functional satisfaction score to ensure the professionalism and rigor of the design, and further improve the performance and user satisfaction of the user operation layer. Through the above strategies, this study fully exploits the advantages of digital twin technology at the user operation layer, while overcoming some challenges to achieve an efficient, reliable, and user-friendly visual management module; To quantitatively evaluate the performance of the user operation layer, a performance scoring formula, UOP, was introduced. By comprehensively considering the weights of various indicators and the functional satisfaction score, we ensure the professionalism and rigor of the design, further improving the performance and user satisfaction of the user operation layer. in is the performance score of the user operation layer, is the number of evaluation indicators, The weight of the indicator, It is a functional satisfaction score used to quantify the performance of the user operation layer and ensure the professionalism and rigor of the design.

[0021] Business logic layer design: The business logic layer design of the intelligent auxiliary control system for new energy power plants is the core of the system's functionality, and its capabilities are enhanced through digital twin technology. It processes requests received from the user operation layer, executes the corresponding business logic, and interacts with the data storage layer to obtain or update the required data. The design of this layer directly affects the system's processing efficiency, user experience, and system stability. In the development of the business logic layer, a modular design approach was adopted. Incorporating digital twin technology, the business logic was divided into multiple independent modules, such as data analysis, fault diagnosis, equipment management, and alarm processing. These modules synchronize real-time data through the digital twin model to provide accurate operation and maintenance decision support. To ensure efficient and reliable operation of the business logic layer, we attach importance to the following design aspects: Data processing efficiency: using digital twin technology to optimize algorithms and data structures, improve data processing speed and accuracy, and ensure that the system can quickly respond to user requests and execute business logic; Exception handling mechanism: A well-designed exception handling mechanism predicts and identifies potential faults through the digital twin model, captures and handles various abnormal situations in a timely manner, ensures the stable operation of the system, and provides users with clear fault diagnosis results; The versatility and reusability of business logic. Considering this versatility and reusability, digital twin technology can be used to reuse the same or similar business logic in different scenarios, reducing development workload and improving development efficiency. In addition, the business logic layer is closely integrated with the data storage layer to design an efficient data access interface. Through the caching mechanism, frequent access to the database is reduced, and digital twin technology is used to further improve system performance. The design of the business logic layer closely follows the overall system architecture and business requirements. The business logic processing flow is continuously optimized and adjusted to adapt to the complex scenarios of new energy power station operation and maintenance, ultimately achieving an efficient and stable intelligent auxiliary control system. The performance of the business logic layer is continuously evaluated and optimized to ensure that it plays a key role in the system and meets the various business needs of users. Performance evaluation and optimization of the business logic layer is an ongoing process. By collecting user feedback, monitoring system operation status, and conducting performance tests, the business logic layer is continuously adjusted and optimized to ensure its efficient operation in the system and provide stable and reliable business processing capabilities. The performance evaluation of the business logic layer uses the formula: in is the performance score of the business logic layer, is the number of evaluation indicators, The weight of the indicator, It is the functional satisfaction score. This formula ensures the professionalism and rigor of the business logic layer design by determining the evaluation indicators and their weights, collecting the satisfaction scores of each indicator, calculating the weighted scores and summing them.

[0022] Data storage layer design: In the intelligent auxiliary control system of new energy power plants, the key to the design of the data storage layer is to ensure the security, stability and efficient management of all operational data. It provides data support for the system's business logic layer and includes various types of data such as real-time monitoring data, historical data, system configuration information, and user information. Therefore, the design of this layer directly affects the performance and reliability of the entire system. We chose to design the data storage layer by combining relational and NoSQL databases. Relational databases are used to store structured data, such as user information and system settings. Their transactional nature, consistency, and relational integrity are crucial for stable system operation. NoSQL databases, on the other hand, are used to store semi-structured or unstructured data, such as real-time monitoring data and historical data. They offer higher write and query performance and are more efficient when processing large amounts of data. To ensure high performance and high availability of the data storage layer, the following key strategies were implemented: Data sharding and load balancing: Data is distributed across multiple storage nodes through data sharding. Combined with load balancing technology, this improves data read and write capabilities and ensures efficient data operations. Data backup and recovery: regularly back up data and ensure the safe storage of backup data so that data can be quickly restored in the event of data loss or system failure to ensure the continuous operation of the system; Data encryption and secure access: encrypt sensitive data and implement strict data access control to ensure data security and privacy; In addition, the design of the data storage layer also takes into account data consistency and integrity. By implementing transaction management and setting reasonable data integrity constraints, the accuracy and reliability of the data are ensured. Efficient data indexing and query optimization technologies are also used to accelerate data retrieval and improve the response time of user queries. Through the above design and technical implementation, the data storage layer not only meets the data needs of new energy power station operation and maintenance management, but also ensures secure, stable, and efficient data access. This provides solid data support for the stable operation and efficient management of the new energy power station intelligent auxiliary control system. Performance evaluation of the data storage layer is a key step in ensuring that the system design meets business needs. Performance can be quantified using the formula: in Indicates the performance score of the data storage layer, is the number of evaluation metrics, It is The weight of the indicator, It is Response time score for each metric.

[0023] Design of real-scene 3D model of the station: The real-life 3D model of a station is the core application of digital twin technology. Through high-precision data acquisition and intelligent processing technology, it achieves a virtual replica of a new energy power station. Its construction process is divided into five stages: data acquisition, processing, modeling, optimization, and verification. Data acquisition combines drone aerial photography with ground laser scanning. The former efficiently acquires large-area terrain images, while the latter generates high-precision point cloud data, which together provide the basis for the model. The data processing stage optimizes data quality through denoising, completion, and classification to ensure the accuracy of subsequent modeling. Model construction is based on geometric feature extraction and gridding technology, converting point clouds into continuous 3D surface models, and improving details and visual effects through optimization technology, such as grid simplification to improve computational efficiency and texture mapping to enhance realism. The verification stage ensures that the model is consistent with the status and operating logic of the actual station through dynamic data updates and error analysis. In the model rendering and display process, ray tracing, GPU acceleration and other technologies are used to achieve efficient and realistic three-dimensional rendering, and Web technology is combined to develop a user-friendly interactive interface; the display interface intuitively presents real-time equipment data and warning information through charts and dashboards, supports multi-terminal operation and immersive experience, and helps operation and maintenance personnel remotely monitor and make decisions; this technology system not only provides a high-precision digital twin platform for new energy power stations, but also plays a key role in intelligent management, fault diagnosis and performance optimization, promoting the transformation of power station operation and maintenance towards digitalization and intelligence.

[0024] Statistical analysis model and algorithm design: Data analysis for new energy power plants begins with data preprocessing and feature extraction. To address noise and missing values ​​in power plant operating data, statistical methods are used to clean outliers and fill in missing data to ensure data integrity. Statistical features (such as mean and extreme values) and frequency domain features are then extracted from time series. Key indicators are then screened using physical models. New features are constructed through arithmetic operations (temperature difference, humidity ratio) and function transformations (periodic and exponential functions), enhancing the dataset's information content and predictive potential. In data mining and machine learning applications, neural network models are trained based on the distribution patterns and correlation analysis of multidimensional numerical data (such as temperature and power). The model captures complex nonlinear relationships through an adaptive learning rate optimization algorithm to predict energy output. The evaluation phase uses confusion matrices, classification reports, and visualization tools (such as ROC curves) to verify model performance and confirm the model's effectiveness in analyzing power plant operating efficiency and environmental adaptability. This technology system provides core support for the intelligent auxiliary control system of new energy power stations through a data-driven approach, helping to optimize operational decisions and fault warnings, and promoting the digital and intelligent upgrade of power station management.

[0025] Data baseboard design: Through the analysis of business entities and their relationships, the data base structure of this system was constructed, and the system uses MySQL as the data management platform; Power station basic information entity (t_station) The power station basic information entity is mainly used to store basic information of new energy power stations. The main fields include: power station ID, power station name, location, type, and capacity. This table is one of the core tables in the power station intelligent auxiliary control system database and is used to associate various equipment and monitoring data in the power station. Power station information table (t_station) Column Name Data Type length Allow Empty Is it a primary key? illustrate StationID int 10 no yes Unique identifier of the power station StationName varchar 100 no no Name of the power station Location varchar 255 no no Geographical location of the power station Type varchar 50 no no Power station type (such as wind power, solar power, charging pile) Capacity decimal 20,2 yes no The maximum generating capacity of the power station, in megawatts (MW) Equipment asset information entity (t_equipment) The Equipment Asset Information entity is used to record detailed information about all equipment in the power plant. The main fields include: equipment ID, equipment type, model, installation date, and status. This table provides basic data for equipment maintenance, monitoring, and fault analysis. Equipment Information Table (t_equipment) Column Name Data Type length Allow Empty Is it a primary key? illustrate EquipmentID int 10 no yes Unique identifier for the device Type varchar 50 no no Device Type Model varchar 50 no no Device Model InstallDate date no no Installation Date Status varchar 20 no no Current status (e.g., operation, maintenance) Device monitoring data entity (t_monitoringdata) The device monitoring data entity is used to store data collected from various sensors and monitoring devices. The main fields include: data ID, device ID, timestamp, data type, and monitoring value. This data is crucial for analyzing device performance and environmental conditions. Monitoring data table (t_monitoringdata) Column Name Data Type length Allow Empty Is it a primary key? illustrate DataID int 10 no yes Unique identifier for monitoring data EquipmentID int 10 no no Associated device identifier Timestamp datetime no no Timestamp of data records DataType varchar 50 no no Data type (e.g., temperature, current) Value decimal 10,2 no no Monitoring value Equipment fault record entity (t_faultrecord) The Equipment Fault Record entity is used to record all fault information that occurs in power plant equipment. The main fields include: fault ID, equipment ID, occurrence time, fault type, and resolution status. This table is an important basis for fault response and subsequent analysis.

[0026] Fault record table (t_faultrecord) Column Name Data Type length Allow Empty Is it a primary key? illustrate FaultID int 10 no yes Unique identifier of the fault record EquipmentID int 10 no no Associated device identifier OccurrenceTime datetime no no Failure time FaultType varchar 50 no no Fault type ResolutionStatus varchar 20 no no Resolution status (resolved, unresolved) Equipment Maintenance Record Entity (t_maintenancerecord) The Equipment Maintenance Record entity is used to record detailed information about equipment maintenance activities. The main fields include: Maintenance ID, Equipment ID, Maintenance Date, Maintenance Type, and Maintenance Personnel. This table helps track maintenance history and plan future maintenance work.

[0027] Maintenance record table (t_maintenancerecord) Column Name Data Type length Allow Empty Is it a primary key? illustrate MaintenanceID int 10 no yes Maintain a unique identifier for the record EquipmentID int 10 no no Associated device identifier MaintenanceDate date no no Maintenance Date MaintenanceType varchar 50 no no Maintenance Type MaintenancePersonnel varchar 50 no no Personnel responsible for maintenance System user information entity (t_user) The system user information entity is used to store the personal information of system users. The main fields include: user ID, user name, password, and user role. It is used for user authentication and authorization and is the basis of system security management. User information table (t_user) Column Name Data Type length Allow Empty Is it a primary key? illustrate UserID int 10 no yes Unique identifier for the user Username varchar 50 no no username Password varchar 100 no no password Role varchar 20 no no User Roles Visualization platform implementation During the development of the intelligent auxiliary control system for new energy power plants, we closely followed the design principles and architectural planning to achieve an efficient, reliable, and user-friendly system. This included specific implementation methods for the user operation layer, business logic layer, and data storage layer. User operation layer implementation: The user operation layer is the frontier of user interaction for the new energy power station intelligent auxiliary control system. This layer is built using the React framework and integrates digital twin technology to ensure efficient page rendering and real-time data interaction. React's component-based architecture allows for the simulation of digital twin views, making the development process more efficient and modular, and facilitating subsequent maintenance and upgrades. Environment setup: The command-line tool create-react-app was used to quickly build the project infrastructure to support the digital twin view: npx create-react-app new-energy-station-control-system cd new-energy-station-control-system npm start Interface layout: Use React's JSX syntax to define the main layout of the application, including the navigation bar, sidebar, and content area; quickly implement beautiful and responsive digital twin interface design; Data interaction: Data interaction with the backend is carried out through React components, the digital twin model is updated in real time, and real-time data and status monitoring of the system are provided; Security considerations: To improve security, JWT-based authentication is implemented on the front end. After a user successfully logs in, the server returns a JWT, which the front end stores in localStorage and appends to each subsequent request to implement authentication and permission control. Through the above technologies, the user operation layer not only provides a rich interface and smooth user experience visually, but also meets the requirements of the intelligent auxiliary control system of new energy power stations in terms of data interaction and security. Business logic layer implementation: The business logic layer is a key component in the intelligent auxiliary control system of new energy power plants, handling core business operations. It not only responds to requests from the user operation layer and executes necessary business logic, but also interacts with the data storage layer to read or update data. By integrating digital twin technology, the business logic layer can simulate and predict system behavior and optimize decision-making processes. To achieve an efficient and maintainable business logic layer, the Spring Boot framework was adopted to support rapid development and easy deployment. Spring Boot project setup: I quickly generated a Spring Boot project framework using Spring Initializr (https: / / start.spring.io / ), selecting dependencies such as Web, JPA, Security, and MySQL, providing a robust foundation for the application. # Example: Creating a new project via Spring CLI spring init --dependencies=web,data-jpa,security,mysql new-energy-station-control-system cd new-energy-station-control-system Define a RESTful API: In the business logic layer, define a RESTful API to handle user requests. Use the @Controller or @RestController annotation to create multiple controller classes, each responsible for a set of related business logic processing. Data Interaction and Security: For data interaction, we leveraged Spring Data JPA to interact with the MySQL database. By defining the Repository interface, we simplified the data access layer code. We also implemented JWT-based user authentication and authorization using the Spring Security framework, increasing application security. Through the above design and implementation, the business logic layer can not only efficiently process requests from the user operation layer and execute complex business logic, but also ensure data consistency and integrity through interaction with the data storage layer. In addition, by implementing a JWT-based authentication and authorization mechanism, the system security is enhanced, ensuring secure access and operation of data. Data storage layer implementation: The data storage layer is the core component responsible for data persistence in the intelligent auxiliary control system of new energy power plants. It not only needs to ensure stable data storage but also provide efficient data access and processing capabilities. To this end, MySQL and Spring Data JPA are used to store relational data, and Redis is introduced as a cache database to improve data access efficiency. MySQL database configuration: Configure the MySQL database connection information in the application.properties file to ensure that the Spring Boot application can successfully connect to the MySQL database; Entity class and Repository definition: To operate the data in the database, the entity class (Entity) is defined to map the database table, and the Repository interface of Spring Data JPA is used to simplify the data access layer (DAL) code; Redis cache configuration: For frequently accessed data, performance can be improved by introducing Redis as a cache layer. Configuring Redis in Spring Boot is relatively simple. You only need to add the Redis connection information to application.properties and define a configuration class to enable cache support. Data security and backup: To ensure data security, in addition to implementing access control and encryption at the database level, databases are backed up regularly to prevent data loss or corruption. This process can be automated through scripting or performed manually using tools provided by the database management system. Through the above configuration and implementation steps, the data storage layer not only achieves stable data storage, but also significantly improves data processing efficiency by introducing Redis cache. In addition, it attaches great importance to data security and backup to ensure stable system operation and data security. This design and implementation strategy ensures that the data storage layer can efficiently and stably support the operation of the intelligent auxiliary control system of the new energy power station, providing a solid data foundation for the upper-level business logic.

[0028] Realization of the real-scene 3D model of the station: Model scanning and creation: The construction of a three-dimensional model of the actual site is the core link of the digital twin system of the new energy power station. Through high-precision data acquisition and processing technology, the scanning and creation process of the digital reproduction model of the physical site is realized. As shown in the figure, its architecture is divided into five key stages: data acquisition, data processing, model construction, model optimization and model verification. In the data acquisition stage, UAV aerial photography and ground laser scanning technology are used to obtain two-dimensional images and three-dimensional point cloud data of the site respectively, ensuring the complete geometric and texture information of the terrain, buildings and equipment. In the data processing stage, the original data is cleaned and structured through denoising, completion and classification algorithms to lay the foundation for model construction. In the model construction stage, based on the processed point cloud data, a preliminary three-dimensional mesh model is generated through geometric feature extraction and triangulation algorithms, and then texture mapping and coloring are used to enhance the visual realism. In the optimization stage, mesh simplification, smoothing and subdivision technologies are used to balance model accuracy and performance. Finally, the geometric consistency and functional reliability of the model are verified through dynamic data synchronization and error analysis to ensure that it can accurately map the actual site scene and support subsequent simulation analysis. Model rendering and display: The model rendering and display phase focuses on transforming 3D models into user-interactive visual interfaces. Rendering technology simulates real-world light interactions based on physical lighting models (such as ray tracing and photon mapping), combined with GPU acceleration to improve rendering efficiency and generate high-fidelity images. The display interface design emphasizes layered information presentation and user-friendly interaction. Front-end technologies such as WebGL and Three.js enable web-based embedding of models, supporting basic operations like rotation and scaling, and integrating real-time data dashboards and early warning systems. The site's real-life 3D model architecture not only realizes the digital twin of the physical environment, but also establishes a closed loop from data to decision-making through efficient rendering and interactive design. Real-time synchronization between the model and IoT data ensures the accuracy of dynamic updates, while the multi-terminal adaptive display solution meets the operation and maintenance needs in different scenarios. This system provides a visual foundation for intelligent monitoring, fault diagnosis, and performance optimization of new energy power stations, significantly improving management efficiency and system reliability. It is an important practice for the implementation of digital twin technology in the energy field.

[0029] Statistical analysis model and algorithm implementation: Data preprocessing and feature extraction: Statistical analysis of new energy power plants begins with data preprocessing and feature extraction, which are key steps in ensuring data quality and model effectiveness. Raw data often contains noise, missing values, and outliers, requiring cleaning through standardization methods (such as Z-score anomaly detection) and interpolation techniques (such as mean filling). Feature extraction focuses on extracting statistical features (mean, standard deviation) and frequency domain features (Fourier transform) from time series data (such as wind speed and power generation), and combining them with physical models to generate derived features (such as temperature difference and humidity ratio). Feature selection uses filtering or correlation coefficient analysis to screen high-value features, ultimately forming a streamlined and information-rich feature set, laying the foundation for subsequent modeling. Data mining and machine learning applications: The construction of data mining and machine learning models requires comprehensive consideration of data sources, types, dimensions, and distribution characteristics. For numerical data from new energy power plants (such as environmental parameters and power output), a neural network model is used for nonlinear modeling. Its structure consists of an input layer (corresponding to feature dimensions), a hidden layer (with a sigmoid activation function), and an output layer (with linear activation). The training process optimizes weights using a backpropagation algorithm, uses the mean squared error (MSE) as the loss function, and uses the Adam optimizer to adaptively adjust the learning rate. The model enhances the expressiveness of input information through feature construction (such as using trigonometric functions to process wind direction and exponential functions to process temperature), thereby improving prediction accuracy. Model evaluation and optimization: Model performance is evaluated using metrics such as confusion matrices, classification reports, and receiver operating characteristic (ROC) curves. Further optimization can be achieved by adjusting the number of network layers and neurons or introducing regularization methods (such as Dropout) to prevent overfitting. Ultimately, the framework achieves a closed-loop process from data cleaning to high-precision prediction, providing reliable theoretical support for power plant operational efficiency analysis and intelligent control.

[0030] Data backplane implementation: The data backplane is implemented based on the business needs of the intelligent auxiliary control system for new energy power stations. It supports the operation of the entire system by constructing six core data entities. The power station basic information entity (t_station), as the core data node of the entire system, stores basic attribute information for all power stations, including key data such as station identification, name, geographic location, type, and installed capacity. This entity primarily serves functional modules such as power station archive management, regional distribution analysis, and capacity statistics. When the system needs to display a list of power stations or filter power stations, it performs data queries based on this entity. The power station ID, as the primary key, not only uniquely identifies each power station within this entity but is also referenced as a foreign key in other entities, such as the equipment asset information entity and the monitoring data entity, establishing relationships between power stations and equipment, and between power stations and monitoring data. Cross-entity queries, such as querying all equipment or monitoring data under a specific power station, are implemented through these relationships. The equipment asset information entity (t_equipment) records the basic information and technical parameters of all equipment in the power plant, including fields such as equipment ID, type, model, installation date, and operating status. This entity primarily serves functions such as equipment asset management, status monitoring, and maintenance planning in the system. When the system needs to display a list of equipment or perform equipment status statistics, it queries the data in this entity. The equipment ID, as the primary key, not only uniquely identifies each device within this entity but is also referenced as a foreign key by the monitoring data entity, fault record entity, and maintenance record entity, establishing relationships between the device and monitoring data, equipment and fault records, and equipment and maintenance records. When the system performs full equipment lifecycle management, it uses these relationships to link the static information of the device with dynamic monitoring data, fault records, and maintenance records to form a complete equipment archive. The monitoring data entity (t_monitoringdata) stores time-series monitoring data collected from various sensors, including fields such as data ID, associated device ID, timestamp, data type, and monitoring value. This entity primarily serves real-time monitoring, historical data query, and performance analysis functions in the system. When displaying real-time monitoring images or generating historical trend curves, the system performs data queries and calculations based on this entity. The data ID serves as the primary key to ensure the uniqueness of each monitoring record, while the device ID serves as the foreign key to ensure that each piece of monitoring data is accurately associated with the corresponding device. When performing equipment performance analysis or fault warnings, the system often needs to perform statistical analysis and pattern recognition based on the historical data in this entity. The fault record entity (t_faultrecord) records all equipment fault events, including fields such as the fault ID, associated device ID, occurrence time, fault type, and resolution status. This entity primarily serves fault management, reliability analysis, and maintenance decision-making within the system. When the system needs to display a fault list or perform fault statistical analysis, it queries the data in this entity. The fault ID, as the primary key, ensures the uniqueness of each fault record, while the device ID, as the foreign key, ensures that each fault record is accurately associated with the corresponding device. When performing equipment reliability analysis, the system often calculates key metrics such as the device's MTBF (mean time between failures) based on the fault records in this entity. The maintenance record entity (t_maintenancerecord) records all maintenance activities for a device, including fields such as the maintenance ID, associated device ID, maintenance date, maintenance type, and maintenance personnel. This entity primarily serves maintenance plan management, maintenance history query, and maintenance effectiveness evaluation within the system. The system queries the data in this entity when displaying maintenance work orders or performing maintenance statistics. The maintenance ID, as the primary key, ensures the uniqueness of each maintenance record, while the device ID, as the foreign key, ensures that each maintenance record is accurately associated with the corresponding device. When formulating preventive maintenance plans, the system often uses the historical maintenance records in this entity to predict the next maintenance date. The user information entity (t_user) stores the account information of all users in the system, including fields such as user ID, user name, password, and role. This entity mainly serves functions such as user authentication, permission control, and operation auditing in the system. When a user logs in to the system, the system queries this entity for identity authentication. When the system needs to check user permissions, it makes a judgment based on the role field in this entity. The user ID serves as the primary key to ensure the uniqueness of each user, while the role field determines the scope of the user's operation permissions in the system. When the system conducts a security audit, it often combines the user information in this entity with the operation logs of other entities for tracing.

[0031] Construction of a 3D model of the actual site based on digital twins: Digital twin technology uses physical models, sensor data, and analytical tools to map the state, behavior, and performance of physical systems in real time in digital space. It enables comprehensive monitoring, simulation, and optimization of physical systems, providing intelligent support for their management and decision-making. The application of digital twin technology in intelligent auxiliary control systems for new energy power plants primarily involves the construction of a 3D model of the actual site, the integration and management of equipment information, and the rendering and display of the model. System visualization technology: In the design and implementation of intelligent auxiliary control systems for new energy power plants, the application of system visualization technology not only improves monitoring efficiency, but more importantly, achieves comprehensive optimization of power plant operation and maintenance management through highly integrated and in-depth analysis. Digital twin technology builds a virtual model of a new energy power plant and synchronizes the status data of physical equipment with the virtual model in real time, enabling comprehensive monitoring and analysis of the power plant. This technology not only includes basic 3D modeling but also covers equipment performance simulation, fault simulation analysis, and other aspects. By communicating in real time with sensors installed on physical equipment, the digital twin model can accurately reflect the equipment's real-time operating status, including but not limited to key parameters such as temperature, pressure, and current. In addition, through deep learning of historical data, the digital twin model can predict the equipment's future operating trends, providing a scientific basis for operation and maintenance decisions. In system visualization technology, the role of data mining and analysis is to extract valuable information from large amounts of historical and real-time data to guide operational decisions. This process involves complex steps such as data preprocessing, feature selection, model building, and optimization. On the one hand, methods such as time series analysis and cluster analysis can identify abnormal patterns in equipment operation and provide early warning of faults. On the other hand, fault diagnosis models based on machine learning algorithms can accurately determine and classify fault types, providing precise guidance for fault handling. In addition, in-depth data analysis can also optimize and analyze the power plant's operating efficiency and energy consumption patterns, further improving the plant's economic benefits. The core of system visualization lies in how to present complex data and models to operation and maintenance personnel in an intuitive and easy-to-understand format. A highly interactive graphical interface design is adopted here, including a dynamic data dashboard, a real-time updated 3D model display, and pop-up information windows for fault warnings. Through these intuitive interfaces, operation and maintenance personnel can quickly grasp the operating status of the power station and respond to warning information in a timely manner. More importantly, the system also supports advanced functions such as scenario simulation and troubleshooting through interactive operations, greatly improving the efficiency and accuracy of operation and maintenance.

[0032] Integration and management of equipment information: Integrating and managing information about equipment within a site is a key step in the application of digital twin models. It ensures that the virtual model not only matches reality in appearance but also accurately reflects the actual status of each piece of equipment in terms of function and performance. This process includes the standardization and integration of equipment information, as well as its real-time updating and management. Standardization of equipment information: Standardization of equipment information is the first step in the integration process. Considering the wide variety of equipment in a new energy power station, including but not limited to wind turbines, solar panels, transformers, etc., each device has its own specific parameters and performance indicators. Therefore, this information needs to be standardized to ensure consistency in the information format to facilitate subsequent integration and analysis. Information standardization can be achieved by defining a unified data model. The standardization of equipment information is described as follows: in, A data model representing a device, including its unique identifier , device type, and a series of Several key parameters. By defining a unified data model, the consistency and compatibility of device information can be ensured, facilitating subsequent integration and analysis; Equipment information integration: Integrating standardized equipment information into the site's 3D model is a critical step. This process typically requires specialized software tools, such as product lifecycle management (PLM) systems and computer-aided design (CAD) software. Equipment information integration includes not only static information (such as equipment specifications and models) but also dynamic information (such as real-time status and performance data). Specifically, the PLM system integrates static equipment information into the 3D model. The PLM system manages equipment information throughout its entire lifecycle, ensuring the integrity and consistency of equipment data. Utilize real-time data acquisition systems to associate dynamic information about equipment with 3D models. Dynamic information integration uses sensor networks and IoT technologies to enable real-time monitoring and updating of equipment status. Real-time update and management of device information: To ensure the accuracy and practicality of the digital twin model, a real-time information update mechanism must be implemented. This mechanism collects real-time operating data through various sensors deployed at key locations on the equipment, including temperature sensors, vibration sensors, and current sensors. The data is then transmitted to the backend server via wireless or wired networks. The collected data is cleaned, filtered, and processed to ensure accuracy and availability. The data is then updated to the digital twin model, enabling real-time updates and synchronization. The data synchronization process includes data verification, comparison, and merging to ensure consistency between the model and the actual device status. The data update process can be simplified using the formula: in, and Respectively represent the device information before and after the update, It is the amount of change in device information during this period; The real-time information update mechanism ensures that device information can be reflected in the digital twin model in a timely manner, improving the real-time and accuracy of the model; effective device information management also involves data security, access control, and backup and recovery; the use of database management systems (DBMS) and cloud storage services can improve the efficiency and reliability of data management while ensuring data security; in terms of data security, encryption technology and security protocols are used to ensure the security of data during transmission and storage; through authentication and access control, unauthorized access and data leakage are prevented; in terms of data backup and recovery, device data is backed up regularly to prevent data loss and damage; cloud storage services are used to provide efficient data backup and recovery mechanisms to ensure the continuous availability of data; in terms of access control, strict access control policies are implemented to manage access and operations to device data according to user permissions; through role management and permission allocation, secure access to data by different users is ensured.

[0033] To ensure the accuracy and practicality of the digital twin model, a time series database-based solution is introduced to manage the real-time data of devices. The time series database can efficiently store and query the historical and real-time data of devices, supporting high-frequency data updates and fast data retrieval. Specifically, the real-time and historical data of devices are stored in the time series database. Whenever the status of a device changes, the relevant data is automatically uploaded and updated to the model. The data update process can be described by the following formula: in, Indicates time The device status, Indicates the amount of change in state. Through continuous updates, the model can reflect the latest status of the device in real time; Utilize the analytical capabilities of the time series database to analyze and process real-time data from equipment; extract valuable information through data aggregation, filtering, and analysis to support operation and maintenance decisions and performance optimization.

[0034] Through the standardization, integration, real-time updating and management of equipment information, the digital twin model can accurately reflect the operating status of the new energy power station, support various applications such as operation and maintenance decision-making, performance optimization and fault diagnosis, thereby significantly improving the operational efficiency and management level of the power station.

[0035] Construction of real-scene 3D model of the station: The construction of a real-world 3D model of a station is the core application of digital twin technology. Through high-precision data acquisition and processing technology, it accurately replicates the new energy power station in the real world. The process is mainly divided into five steps: data acquisition, data processing, model construction, model optimization and verification. The following is a detailed description of each step.

[0036] Data collection: Data collection is the basis for building a 3D model. It mainly uses drone aerial photography and ground laser scanning technology to capture the details of the site's terrain, buildings, and equipment. Drone aerial photography uses drones to capture high-resolution images of the station from the air, generating two-dimensional image data of the terrain and buildings. Equipped with high-precision cameras, drones capture the entire station area from multiple angles, ensuring comprehensive and accurate data. Drone aerial photography is fast and efficient, allowing for the acquisition of large-scale image data in a short period of time, making it particularly suitable for areas with complex terrain or difficult-to-reach areas. Terrestrial laser scanning uses laser scanning equipment to collect three-dimensional point cloud data of the station from the ground; the point cloud data includes the spatial coordinates of each point ( ) and color information ( ), the formula is as follows: Laser scanning can capture the fine structure of buildings, equipment, and terrain, generating high-density point cloud data, which provides the basis for 3D model construction. Laser scanning has the characteristics of high precision and high resolution, and can accurately reflect the subtle features of objects; Data processing: The collected raw data needs to be preprocessed, including denoising, completion, and classification, to ensure the purity and usability of the data and prepare for subsequent modeling. The following details the processing steps; Denoising: By identifying and removing outliers in the point cloud, random and systematic errors in the data are reduced, the quality of the point cloud data is improved, and modeling is more accurate. Common denoising methods include Gaussian smoothing and median filtering. Gaussian smoothing smoothes local noise through convolution operations, while median filtering replaces the value of each data point with the median of all points in its neighborhood to effectively suppress the influence of noise. Data completion: For missing areas in the data, use Kriging interpolation or radial basis function (RBF) interpolation methods to infer the value of the missing point from the data of surrounding points to ensure data integrity; the formula is as follows: in, is the estimated missing point, It is the surrounding A known point, is the corresponding weighting coefficient; data completion uses mathematical models and algorithms to predict and fill in missing data, making the point cloud data more continuous and complete; Point cloud classification and segmentation: Point cloud data is classified and segmented based on point attributes such as reflectance and color for more accurate model reconstruction. Classification and segmentation help distinguish different objects and structures, improving model refinement. Common classification algorithms include rule-based classification, supervised learning, and unsupervised learning. Point cloud segmentation methods include region growing, boundary-based segmentation, and cluster analysis. Refined classification and segmentation play a vital role in the subsequent construction of accurate digital twin models. Model construction: Model construction is based on processed point cloud data. Through two key steps, geometric information extraction and triangulation, the point cloud data is converted into a continuous mesh model. Extracting geometric information is the first step in model construction. It aims to identify and extract key geometric features such as edges, corners, and surface contours by calculating the gradient, normal, and curvature of the point cloud, ensuring that the constructed model can accurately reflect the key structures and features of the original object. Based on the extracted geometric information, the Delaunay triangulation technology is applied to mesh the point cloud data. By maximizing the minimum angle in the triangulated mesh, a set of triangular facets with stable topology and reasonable geometry is generated, thereby effectively constructing a continuous three-dimensional surface model.

[0037] The formula is as follows: in, Represents the final constructed three-dimensional model, Represents point cloud data, Indicates that point cloud data Function to convert to three-dimensional model; Model optimization: In order to improve the level of detail and visual effects of the model, model optimization mainly includes technologies such as mesh simplification, mesh smoothing and mesh subdivision; among them, mesh simplification reduces the complexity of the model and improves rendering speed by reducing the number of triangles; commonly used mesh simplification algorithms include edge collapse, vertex aggregation and multi-resolution analysis; mesh smoothing improves the smoothness of the model by eliminating the jagged effect on the model surface; commonly used mesh smoothing algorithms such as Laplace smoothing adjust the vertex position and reduce the normal difference between adjacent vertices, thereby reducing the high-frequency noise on the surface to enhance the visual effect; in scenes with high requirements for local detail accuracy, mesh subdivision is used to make the model more realistic; by adding more vertices and edges, the spatial resolution of the model is improved, making the model more refined in local areas. Common subdivision methods include Loop subdivision and Catmull-Clark subdivision, which can generate smoother surfaces while maintaining the original shape of the model; Model verification: Model verification aims to ensure that the model not only visually matches the actual site, but also accurately simulates the operating status and operation process of the renewable energy power station. To ensure that the model has good real-time and dynamic response capabilities, the model adopts dynamic data management and automatic update mechanisms to ensure that the real-time data of the site can be reflected in the 3D model in a timely and accurate manner. The formula is as follows: in, and Respectively represent the station model status before and after the update, Indicates time Changes in internal station status; In terms of model accuracy assessment, the analysis is conducted by comparing the deviation between the model prediction and the actual observation data; the formula is as follows: in, The error representing the accuracy of the model, and Respectively represent observed data points and model predictions, is the total number of data points; In summary, through the rigorous data collection and processing, model building and optimization, and verification processes described above, a three-dimensional model of the actual site scene was accurately constructed, laying a solid foundation for the implementation of the intelligent auxiliary control system for new energy power stations, enabling digital twin technology to play a key role in simulation analysis, performance optimization, and fault diagnosis.

[0038] Rendering and display of the model: After successfully building a 3D model of the actual site and integrating equipment information, the next important step is to render and display the model. Through high-quality rendering and an intuitive display interface, the digital twin model can provide users with a virtual environment that is realistic, easy to understand, and easy to operate. This process involves two main steps: efficient model rendering technology and user-friendly display interface design. Efficient model rendering technology: Model rendering is the process of converting a 3D model into a high-quality image or video. This process needs to consider multiple factors such as light, material, and texture. To achieve efficient rendering, the following rendering equation is usually used to simulate the interaction between light and objects: in, From the surface point In the direction The radiance of the upper departure, Yes exist The brightness of the self-illumination in the direction, is the bidirectional reflectance distribution function (BRDF) of the surface, It is the arrival point The incident light brightness, is the cosine of the angle between the incident light direction and the surface normal, integrated over all directions of incident light in the hemisphere; By applying the above rendering equations, we can perform lighting calculations and render visual effects on the model, thereby generating realistic images. In addition, to improve rendering efficiency, the following rendering techniques are often used: Ray tracing technology generates highly realistic images by simulating the propagation path of light. It can accurately calculate the reflection, refraction and shadow effects of light to present realistic lighting effects. Photon Mapping: Photon mapping is a global illumination algorithm that generates realistic lighting effects by tracking the propagation and scattering of photons in a scene. Photon mapping is particularly suitable for simulating complex lighting phenomena such as diffuse reflection and caustics. Utilizing the parallel computing capabilities of graphics processing units (GPUs) can significantly increase rendering speeds; modern rendering engines such as NVIDIA's OptiX and AMD's Radeon ProRender support GPU-accelerated rendering. User-friendly display interface design: The display of the model requires not only visual realism but also a good user interaction experience. To this end, the following three aspects should be focused on in the design of the display page: Effective information presentation: The display interface should clearly display important equipment information and operating status, including real-time data, historical records, and warning information. Visual tools such as charts and dashboards should be used to help users quickly understand and analyze data. The interface design should be concise, intuitive, and easy to operate. Modern web technologies should be used to render 3D models directly in the browser. By building an intuitive user interface (UI) such as buttons, sliders, and menus, users can easily browse the model, view equipment information, and perform simulation operations. Users can interact with the model through the mouse, keyboard, and touch screen to perform operations such as rotation, zooming, and panning. The interactive design should take into account the user's operating habits and experience and provide smooth interactive effects. Virtual models can be viewed and operated in the actual environment to enhance understanding of the equipment and site. This approach not only enhances the user experience, but also helps operation and maintenance personnel have a more intuitive understanding when performing equipment maintenance, fault diagnosis, and other operations, thereby improving the efficiency and accuracy of decision-making. Therefore, this application uses Web technologies such as HTML5, CSS3, and JavaScript to embed the rendering results of the model into the web page, realizing the online browsing and interactive functions of the model; through WebGL and the Three.js library, an efficient 3D rendering engine is built, and a user-friendly interactive interface is developed; HTML5 and CSS3 are used for page layout and style design to achieve responsive design and ensure good display effects on different devices; in addition, the web page provides a variety of display functions and interactive modes, which can realize operations such as rotation, scaling, translation, and switching of the model; users can directly access and operate the 3D model through the browser to obtain device information and operation data.

[0039] Statistical analysis models and algorithms for new energy power plants: Data preprocessing and feature extraction: In intelligent auxiliary control systems for new energy power plants, data preprocessing and feature extraction form the foundation of the data analysis process. New energy power plants, such as wind and solar power plants, generate large amounts of operational data. This data often contains noise, missing values, and inconsistent measurements, directly impacting the quality and accuracy of analysis. Therefore, data preprocessing is an integral part of the analysis process, aiming to clean and organize data through a series of technical means, laying a solid foundation for feature extraction and subsequent analysis. For outlier processing, anomaly detection technology based on statistical methods is usually used, and the Z-score method is used to evaluate the degree of abnormality of data points; the formula is as follows: in, represents the observed value, and are the mean and standard deviation of the data set; by calculating the Z-score value of each data point, outliers that are far from the average level can be identified; for the treatment of missing values, interpolation, filling with the mean or median, etc. may be used to ensure the integrity of the data set; Next, feature extraction involves identifying information useful for the analysis objective from the processed data. The key lies in converting the raw data into a set of features that characterize the data. These features can effectively capture the key information of the data and are crucial for improving the performance of the analysis model. In the scenario of a new energy power station, feature extraction may involve extracting statistical features, frequency domain features, or features based on physical models from time series data. For example, from the time series of wind speed and power generation, statistical features such as mean, standard deviation, maximum, and minimum values ​​can be extracted, as well as frequency domain features extracted through Fourier transform. After feature extraction, feature selection is required to select the most informative feature subset from the extracted features for building a statistical analysis model. Common feature selection techniques include filtering-based methods, wrapper selection, and embedded selection. Specifically for the typical photovoltaic and wind farms studied in this application, wind power data includes time-varying data such as wind speed (vector), impeller speed, gearbox temperature, and vibration spectrum (10-2000Hz). Frequency domain features (such as the characteristic frequency of bearing faults, which is 28-35Hz) need to be extracted through short-time Fourier transform (STFT). Photovoltaic data includes irradiance, module temperature, IV curve, dust accumulation index, etc., and trend features (such as the daily irradiance decay rate) need to be extracted through sliding windows. Through the above data preprocessing and feature extraction process, it is possible to effectively and accurately extract information useful for analysis and decision-making from the large amount of operational data of new energy power plants. This not only lays a solid foundation for in-depth data analysis, but also establishes a solid data foundation for the high-precision modeling and intelligent control of digital twin technology in new energy power plants. This process demonstrates the core role of data preprocessing and feature extraction in new energy statistical analysis and is a key step in achieving high-quality analysis results. Data mining and machine learning applications: The application scenarios of data mining and machine learning in a system are related to the specific problems and goals faced. This includes multiple aspects such as the source, type, dimension, distribution, and relationships between data. For these factors, the following detailed considerations are required: Data sources describe how data is acquired and generated, each with its own unique characteristics and application scenarios. For example, sensors can provide real-time data, while databases are more suitable for acquiring structured data. The internet can collect massive amounts of data, while user-generated information is more personalized. Therefore, it is particularly important to select appropriate acquisition and processing strategies based on the specific source of the data. As for data types, it involves the representation and storage methods of data. Different types of data have different characteristics and uses. For example, numerical data is convenient for mathematical and statistical analysis; categorical data is more suitable for classification and cluster analysis; time data is helpful for sequence and trend analysis; spatial data focuses on geographic and spatial analysis; network data can be used to explore relationships and structures. Choosing appropriate analysis and model building methods based on data type is the key to efficient data processing. The dimension of data reflects the number of attributes and features contained in the data, which influences the choice of data analysis and model building methods. From one-dimensional to multidimensional, the difference in data dimension brings about changes in the scope and depth of analysis and presentation. For example, one-dimensional data is suitable for basic analysis, while multidimensional data can support complex high-dimensional analysis. Examining data distribution reveals the range and dispersion of data. Different distribution patterns guide analytical and forecasting methods. For example, uniformly distributed data is suitable for basic analysis, while normally distributed data is more suitable for standard analysis and forecasting. Skewed, grouped, and clustered distributions each have their own characteristics, corresponding to different analytical and forecasting strategies. Relationships between data express the connections and interactions between data points. The type and degree of these relationships are crucial to the choice of data analysis methods. Whether it is comparison, ranking, or ratio analysis, or trend, correlation, causality, or network analysis, understanding the relationships between data helps to make more accurate predictions and analyses. Through in-depth discussion of the above aspects, we can more accurately identify and apply data mining and machine learning in different scenarios to specific problems and goals, and then select the most appropriate methods and strategies to achieve the desired analysis and processing results; In this application, we explored the mining and machine learning applications of new energy power station data. The experimental process and results involved are described below. The goal is to deepen our understanding of power station operating efficiency and environmental adaptability through data analysis: The data comes from a new energy power station, collected through its sensors and database, and includes key information such as operating status, environmental parameters, and energy output. The dataset analyzed in this application contains 10,000 records, each consisting of 10 feature values ​​and 1 predicted value. This data source ensures the immediacy and structure of the information obtained. Regarding the data type, we focus on numerical data, which means that both features and predicted values ​​are in numerical form, such as temperature, humidity, wind speed, wind direction, and power. This type of data type facilitates quantitative analysis and computational processing. In terms of data dimensions, the data processed in this application is 11-dimensional, including 10 features and 1 predicted value. This dimension setting helps to maintain the comprehensiveness and diversity of the data; The distribution of data shows normal distribution characteristics, indicating that both features and predicted values ​​follow certain regularity and stability, forming a bell-shaped distribution pattern; As for the relationship between data, the analysis shows that there is an obvious correlation and causal relationship between the characteristic values ​​and the predicted values, which provides a reliable basis for subsequent data analysis and prediction.

[0040] In terms of application methods, this application adopts a series of steps and techniques for data mining and machine learning, including data preprocessing, feature extraction, model training and result evaluation; the data preprocessing link is particularly important, involving the cleaning, conversion and normalization of raw data, in order to improve data quality and utilization efficiency.

[0041] During the data cleaning process, we employed both deletion and interpolation strategies. Deletion removed duplicate records from the dataset, avoiding redundancy and bias during data analysis. Missing values ​​were addressed through interpolation, aiming to maintain data integrity and continuity. In particular, mean interpolation was employed, replacing missing values ​​with the mean of the corresponding feature. This approach aims to maintain overall data consistency and balance. Through this series of steps and methods, this application aims to deepen the understanding of the operating efficiency and environmental adaptability of new energy power plants and provide a scientific basis for future energy management and optimization; Data conversion refers to adjusting the format and scope of data to meet the requirements of data analysis and model building. Data conversion methods include standardization, normalization, discretization, and encoding. The appropriate method should be selected based on the characteristics and needs of the data. This application uses standardization and encoding methods to convert data. The specific operations are as follows: Standardization: This application uses a standardization method to convert the data features and predictions into a standard normal distribution with a mean of 0 and a standard deviation of 1, in order to eliminate the dimension and bias of the data and improve the comparability and stability of the data. The standardization method is to subtract the mean of each data value from the data and then divide it by the standard deviation of the data to obtain the standardized value. The formula is as follows: in is the original value, is the mean, is the standard deviation, is the standardized value; Data processing methods are used to convert categorical data into numerical data, thereby facilitating data analysis and model building. In the field of data processing, encoding is not only a fundamental technology but also a key bridge between raw data and efficient analysis. For example, the categorical feature of wind direction, which includes different directions such as east, south, and west, is difficult to directly apply to mathematical models without processing. Through encoding, these directions are converted into numerical values ​​such as "0, 1, 2" or more complex binary representations such as "1000, 0100, 0010", making the data more standardized and easier to process and understand by machine learning algorithms. Feature extraction, as an important part of data preprocessing, extracts information that is of substantial help to the prediction task from large-scale raw data; among the many feature extraction methods, feature selection and feature construction were selected in this study to play a key role; feature selection reduces the dimension of the dataset by identifying the features with the most predictive value, which helps the model avoid overfitting and improves training efficiency; filtering, as one of the feature selection methods used in this study, emphasizes the importance of the statistical properties of features for prediction tasks; for example, variance analysis reveals the distribution differences of each feature in the dataset. By setting 0.1 as the threshold, features with large variations, such as temperature and humidity, are screened out. These features are considered to be particularly important for model prediction due to their high variability in the dataset.

[0042] In contrast, correlation coefficient analysis focuses more on the association between features and the target variable. By calculating the correlation coefficient between each feature and the predicted target and screening according to a set threshold, this study can further refine the understanding of the data and ensure that the features ultimately selected have a direct and significant impact on the prediction results.

[0043] Feature construction goes a step further based on feature selection, performing arithmetic or function operations on existing features to create new features in order to capture the complex relationships and patterns that may exist between the data. This process not only enhances the information content of the data set, but also provides the possibility of building more accurate and robust prediction models. For example, by combining features such as temperature and humidity, it may be found that there are certain interactions between these features that have not been directly observed, which may be crucial for predicting certain specific climate conditions or other events that depend on these variables.

[0044] Arithmetic operations use existing features to generate new features through basic arithmetic operations such as addition, subtraction, multiplication, and division to reflect the changes and proportions of the data. This application uses arithmetic operations to construct features for the data. The specific operations are as follows: Temperature difference refers to the difference between the highest temperature and the lowest temperature in the data, which is used to reflect the temperature change of the data. The formula is: in is the temperature difference, is the maximum temperature, is the lowest temperature; this application uses the temperature difference method to construct features of the data and generate new features, such as temperature differences of 5°C, 10°C, 15°C, etc. Humidity ratio refers to the ratio of relative humidity to absolute humidity in the data, which is used to reflect the humidity ratio of the data. The formula is: in is the humidity ratio, is the relative humidity, is the absolute humidity; This application uses the humidity ratio method to construct features on the data and generate new features, such as humidity ratios of 0.5, 0.6, 0.7, etc. Function operation refers to the use of existing features to generate new features through advanced function operations such as trigonometric functions, exponential functions, and logarithmic functions to reflect the nonlinearity and complexity of the data; This application uses the function operation method to construct the characteristics of the data. The specific operations are as follows: New features are generated through basic trigonometric functions such as sine, cosine, and tangent to reflect the periodicity and fluctuation of the data. This application uses trigonometric functions to construct features for the data. The specific operations are as follows: Wind direction sine refers to using the characteristics of wind direction to generate new features through the sine function to reflect the periodic changes in wind direction. The formula is: in is the sine of wind direction, is the wind direction in degrees; the wind direction sine method is used to construct features of the data and generate new features, such as wind direction sine of 0.5, 0.7, 0.9, etc. Wind direction cosine refers to the use of wind direction characteristics to generate new features through the cosine function to reflect the periodic changes in wind direction. The formula is: in is the wind direction cosine, is the wind direction in degrees; this application uses the wind direction cosine method to construct features of the data and generate new features, such as wind direction cosine of 0.5, 0.7, 0.9, etc. Generate new features through exponential, logarithmic, idempotent and other advanced exponential functions to reflect the exponential growth or decay of data. This application uses the exponential function method to construct features for the data. The specific operations are as follows: The temperature index uses the characteristics of temperature to generate new features through the exponential function to reflect the exponential growth of temperature. Its formula is: in is the temperature index, is the temperature in °C. This application uses the temperature index method to construct features of the data and generate new features, such as temperature index of 2.7, 7.4, 20.1, etc. The humidity logarithm uses the characteristics of humidity to generate new features through the logarithmic function to reflect the logarithmic attenuation of humidity. The formula is: in is the logarithm of humidity, is humidity, in %; This application uses the humidity logarithm method to construct features of the data and generate new features, such as humidity logarithm -0.3, -0.1, 0.1, etc.; Model training uses training data to build a model that can predict data through certain algorithms and parameters to achieve data understanding and prediction. Model training methods include linear regression, logistic regression, decision trees, support vector machines, neural networks, etc., and it is necessary to choose an appropriate method based on the characteristics and needs of the data. This application uses the neural network method to train the model on the data. The specific operations are as follows: A neural network is a model that simulates the human nervous system. It consists of multiple input layers, hidden layers, and output layers. Each layer is composed of multiple neurons, each with weights and biases. Through activation functions and loss functions, it achieves nonlinear mapping and optimization of data. The neural network method refers to using the structure and parameters of the neural network to train the data model and generate a neural network model that can predict the data, so as to achieve data understanding and prediction. In a neural network model, in addition to the input layer and output layer, there can also be multiple hidden layers. Each hidden layer consists of several neurons, and the neurons are connected through weight matrices and bias vectors. The output of each neuron is determined by its input, weight, bias, and activation function. The formula is: in, Indicates the Layer The output of a neuron, represents the activation function, Indicates the connection Layer neurons and Layer The weight of a neuron, represents the bias of the j-th neuron in the l-th layer, Indicates the The number of neurons in the layer; The training goal of a neural network model is to minimize the error between the network output and the desired output by adjusting the weights and biases. The commonly used error function is the mean square error function: in, represents the error function, The output layer The output of a neuron, represents the expected output, Indicates the number of neurons in the output layer; In order to minimize the error function, gradient descent or other optimization algorithms can be used to update weights and biases. The basic idea of ​​gradient descent is to update weights and biases with a certain step size along the negative gradient direction of the error function until a local minimum is reached or a certain stopping condition is met. The formula of gradient descent is: in, represents the learning rate, and represents the partial derivative of the error function with respect to weights and biases; In order to calculate the partial derivatives of the error function with respect to the weights and biases, the backpropagation algorithm can be used. Starting from the output layer, the error term of each neuron is calculated layer by layer, and the gradient of each weight and bias is obtained according to the chain rule. The formula of the backpropagation algorithm is: in, Indicates the Layer The error term of each neuron, represents the derivative of the activation function.

[0045] Although the embodiments of the present invention have been shown and described, as detailed above, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the invention, and the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent auxiliary control system for a new energy power station based on digital twins, characterized by: include The data collection layer collects physical entity data of the new energy power station, including equipment operation data, environmental data, and power grid data. The equipment operation data is collected by sensors deployed on the equipment, the environmental data is collected by environmental monitoring equipment, and the power grid data is collected by grid-side sensors. The data processing layer performs preprocessing, feature extraction, and data fusion on the collected data. The preprocessing includes denoising, missing value filling, and outlier processing. The feature extraction includes statistical features, frequency domain features, and physical model-based feature extraction. The data fusion standardizes multi-source heterogeneous data and organizes and stores them by subject and dimension. The digital twin model layer builds a digital twin model of the new energy power station, including a real-world 3D model of the station and an equipment information model. The real-world 3D model of the station is constructed using data collected by drone aerial photography and ground laser scanning technology, and the equipment information model integrates static and dynamic information of the equipment. The intelligent analysis layer implements operating status analysis, fault diagnosis, power generation forecasting, and optimized scheduling based on the digital twin model and data analysis algorithms. The data analysis algorithms include data mining algorithms and machine learning algorithms. The machine learning algorithms are used to build power generation forecasting models and fault diagnosis models. The application service layer provides users with a visual interface and interaction, including 3D visualization scenes and data dashboards, enabling remote monitoring, fault warning, maintenance decision-making, and optimized scheduling, and supports multi-terminal access. The edge computing layer, deployed in edge monitoring units, completes communication access, edge analysis, and data aggregation. Edge gateways are deployed on wind turbine towers and photovoltaic combiner boxes, supporting MQTT and CoAP protocols to collect sensor data in real time and perform local preprocessing. The data backplane layer builds the system's data entity-relationship model, including power plant basic information entities, equipment asset information entities, equipment monitoring data entities, equipment fault record entities, equipment maintenance record entities, and system user information entities. It uses a relational database to implement structured storage and associative query of data. The model optimization module optimizes the digital twin model, including mesh simplification, physical field coupling, and model rendering optimization. It reduces the number of model faces through mesh simplification, achieves real-time physical mapping of the device's operating status through physical field coupling, and enables lightweight loading and cross-platform display of the model through WebGL technology. The predictive maintenance module generates a predictive maintenance plan based on an equipment health assessment model and a remaining life prediction model. The equipment health assessment model integrates multi-parameter data to generate an equipment health index. The remaining life prediction model predicts the remaining life of the equipment based on the Weibull distribution and optimizes spare parts management and maintenance strategies.

2. The intelligent auxiliary control system for a new energy power station based on digital twin according to claim 1 is characterized by: It also includes an information infrastructure layer, which is responsible for data collection, transmission, storage and calculation, including power plant control systems, computing resources, network facilities and data perception networks. The power plant control system is responsible for the daily operation and management of the power plant. Computing resources provide high-performance computing capabilities and support complex data analysis and model simulation. Network facilities ensure real-time transmission of data. The data perception network covers power generation, transmission, distribution, transformation and energy storage, forming a complete data collection system.

3. The intelligent auxiliary control system for a new energy power station based on digital twin according to claim 1 is characterized by: It also includes a digital twin technology layer, which enables intelligent management of power plants through a data-driven approach. It includes three modules: models and algorithms, visualization models, and data backplane. The model and algorithm module uses data analysis, power generation forecasting, and fault prediction to conduct real-time monitoring of the power plant's operating status and trend forecasting. The visualization model module transforms complex power plant operating data into an intuitive graphical interface through three-dimensional modeling and multi-dimensional data display. The data base module is responsible for data aggregation, governance and mining, while providing stable data support for upper-level models.

4. The intelligent auxiliary control system for a new energy power station based on digital twin according to claim 1 is characterized by: It also includes a user operation layer, an interface for direct interaction between the system and the user, which provides a graphical user interface for operators to perform monitoring, control, and management tasks. It uses modern Web technology, supports access to the system from a variety of terminal devices, and can also display the digital twin model of the power plant in a graphical manner.

5. The intelligent auxiliary control system for a new energy power station based on digital twin according to claim 1 is characterized in that: It also includes a data storage layer, which is responsible for the storage and management of all data in the system, including real-time monitoring data, historical data, system configuration and user information. It uses a relational database management system or NoSQL database to achieve efficient storage, fast retrieval and secure access to data. It is responsible for providing data support for the digital twin model, including real-time data, historical data, model parameters and simulation results.

6. The intelligent auxiliary control system for a new energy power station based on digital twin according to claim 4 or 5, characterized in that: It also includes a business logic layer, which processes requests from the user operation layer and performs specific business logic processing. It is responsible for interacting with the data storage layer to obtain or update stored data. By using digital twin technology, it connects physical entities with their digital representations and uses physical models, sensor data and historical data to build a virtual model of the new energy power station.

7. The intelligent auxiliary control system for a new energy power station based on digital twin according to claim 1 is characterized by: The process of constructing the real-scene three-dimensional model of the station includes five stages: data collection, processing, modeling, optimization and verification.

8. The intelligent auxiliary control system for a new energy power station based on digital twin according to claim 1 is characterized in that: It also includes statistical analysis of new energy power stations, which is achieved through statistical analysis models and algorithms.

Images