Wind power generation intelligent interaction integration system based on large language model
By integrating a wind power generation intelligent interactive system based on a large language model, integrating view, scene construction, rendering and knowledge graph components, and combining with a robot assistant, the system solves the problem of low efficiency in interactive operation in existing technologies, and achieves efficient and diversified user interaction and system optimization.
Patent Information
- Application Number
- CN202411250627.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-09-06
AI Technical Summary
Existing intelligent interaction technologies lack integration in the fields of virtual reality, robotics, and automation systems, resulting in the inability to achieve efficient interactive operations and meet diverse user needs.
A wind power generation intelligent interactive integration system based on a large language model is adopted, including view components, scene building components, rendering components, large language model components and knowledge graph components. It provides visualized scenes through rendering and stylized view fusion, and combines robot assistant components to realize the response to interactive requests and environmental data collection, thereby optimizing the overall performance of the system.
It improves the visual experience and interactive efficiency, meets diverse user needs, enhances the overall performance and execution capabilities of the system, and reduces the risks and costs of manual operation.
Smart Images

Figure CN119025549A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of intelligent interaction technology, and in particular to a wind power generation intelligent interaction integrated system based on a large language model. Background Art
[0002] With the rapid development of intelligent technology, interactive environments and intelligent systems are becoming increasingly important in the fields of virtual reality, robotics, and automation systems. However, existing intelligent interactive technologies often lack integration and have limited functions, which makes it impossible to achieve efficient interactive operations and difficult to meet the diverse needs of users. Summary of the invention
[0003] The disclosed embodiment provides a wind power generation intelligent interactive integration system based on a large language model to solve the problem of being unable to achieve efficient interactive operations.
[0004] The disclosed embodiment provides a wind power generation intelligent interactive integration system based on a large language model, including: a view component, a scene construction component, a rendering component, a large language model component and a knowledge graph component; The rendering component is connected to the scene construction component and is used to render the input view of the target user to obtain a rendered new view; The scene construction component is connected to the view component and is used to stylize the input view to obtain a stylized view, and merge the stylized view with the rendered new view to obtain a scene view; The view component is used to display the scene view; The view component is connected to the large language model component and is used to receive interaction requests from target users; A large language model component, connected to the knowledge graph component, is used to respond to the interaction request and determine the response data corresponding to the interaction request based on the data in the knowledge graph component; View components are also used to display response data.
[0005] In an exemplary embodiment of the present disclosure, a large language model-based wind power generation intelligent interactive integrated system further includes a robotic assistant component; The robot assistant component is connected to the large language model component, and is used to respond to the command of the target user, perform retrieval based on the large language model component, determine the control instruction corresponding to the command based on the retrieval result, and control the external robot to perform the action corresponding to the control instruction according to the control instruction.
[0006] In an exemplary embodiment of the present disclosure, the robot assistant component is connected to the knowledge graph component; The robot assistant component is also used to collect environmental data and send the environmental data to the knowledge graph component; The knowledge graph component is used to update the knowledge base based on environmental data.
[0007] In an exemplary embodiment of the present disclosure, a wind power generation intelligent interactive integration system based on a large language model further includes a wind power generation scenario component; The wind power scenario component is connected to the large language model component; The view component is also used to receive a first request from a target user; The large language model component is also used to determine wind power generation scenario data from the wind power generation scenario component in response to the first request; The view component is also used to display wind power generation scenario data.
[0008] In an exemplary embodiment of the present disclosure, a wind power generation scenario component includes a wind turbine, a wind sensor, a control module, a data monitoring module, and a maintenance module; The wind turbine is connected to a wind sensor, the wind sensor is connected to a control module, the control module is connected to a data monitoring module and a maintenance module respectively; the control module, the data monitoring module, and the maintenance module are all connected to a large language model component; The wind sensor is used to collect wind condition data and send the wind condition data to the control module; The data monitoring module is used to monitor the operation data, and the maintenance module is used to manage the maintenance data; Among them, wind power generation scenario data includes operation data or maintenance data.
[0009] In an exemplary embodiment of the present disclosure, a wind power generation intelligent interactive integration system based on a large language model further includes: The view component is also used to receive a second request from the target user; The large language model component is further used to determine the target scenario based on the scenario building component in response to the second request; The view component is also used to display the target scene.
[0010] In an exemplary embodiment of the present disclosure, a wind power generation intelligent interactive integration system based on a large language model further includes: The view component is also used to receive a third request from the target user; The large language model component is further used to update the scene data based on the scene component in response to a third request; The view component is also used to display updated scene data.
[0011] In an exemplary embodiment of the present disclosure, the rendering component is specifically used to render the input view based on a 3D Gaussian sputtering algorithm to obtain a rendered new view.
[0012] In an exemplary embodiment of the present disclosure, the knowledge graph component includes a knowledge base, which includes external data, historical data, and industry data; The large language model component is connected to the knowledge graph component through the knowledge graph interface.
[0013] In an exemplary embodiment of the present disclosure, the view component is also used to display and render a new view.
[0014] The beneficial effects of a wind power generation intelligent interactive integration system based on a large language model provided by the embodiment of the present disclosure are: The disclosed embodiment provides a visual scene view for the target user through the rendering of the input view by the rendering component, and the fusion of the stylized view and the rendered new view by the scene construction component, which greatly enhances the visual experience. Secondly, the connection between the view component and the large language model component enables the target user's interaction request to be responded to in a timely manner. With the help of the rich data in the knowledge graph component, the large language model component can accurately determine the response data corresponding to the interaction request, thereby achieving efficient interactive operations and meeting the diverse needs of users. Therefore, the close collaboration and efficient integration between the components optimize the overall performance of the system and greatly improve the efficiency of interactive operations. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0016] Figure 1 It is a structural schematic diagram of a wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure; Figure 2 is a structural schematic diagram of another wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure; Figure 3 is a use case diagram of a wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure; Figure 4 is a component diagram of a wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure; Figure 5 is a sequence diagram of a wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure; Figure 6 is an activity diagram of a wind power generation intelligent interactive integrated system based on a large language model provided by an embodiment of the present disclosure; Figure 7 It is a state diagram of a wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure. DETAILED DESCRIPTION
[0017] In order to enable people in the technical field to better understand the present solution, the technical solution in the embodiment of the present solution will be clearly described below in conjunction with the drawings in the embodiment of the present solution. Obviously, the described embodiment is an embodiment of a part of the present solution, not all of the embodiments. Based on the embodiments in the present solution, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present solution.
[0018] The term "including" and any other variations in the specification and claims of this solution and the above drawings mean "including but not limited to", and is intended to cover non-exclusive inclusions and is not limited to the examples listed in the text. In addition, the terms "first" and "second" are used to distinguish different objects, not to describe a specific order.
[0019] The following is a detailed description of the implementation of the present disclosure in conjunction with the specific drawings: Figure 1 A schematic structural diagram of a wind power generation intelligent interactive integration system based on a large language model provided in an embodiment of the present disclosure. Figure 3 It is a use case diagram of a wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure. Figure 4 is a component diagram of a wind power generation intelligent interactive integration system based on a large language model provided by an embodiment of the present disclosure. Figure 1 , Figure 3 and Figure 4 The wind power generation intelligent interactive integration system based on the large language model includes: View component, scene building component, rendering component, large language model component and knowledge graph component; The rendering component is connected to the scene construction component and is used to render the input view of the target user to obtain a rendered new view; The scene construction component is connected to the view component and is used to stylize the input view to obtain a stylized view, and merge the stylized view with the rendered new view to obtain a scene view; The view component is used to display the scene view; The view component is connected to the large language model component and is used to receive interaction requests from target users; A large language model component, connected to the knowledge graph component, is used to respond to the interaction request and determine the response data corresponding to the interaction request based on the data in the knowledge graph component; View components are also used to display response data.
[0020] In this embodiment, target users include but are not limited to users and managers.
[0021] The rendering component is responsible for rendering the input view of the target user to obtain a rendered new view. For example, by using various graphics technologies and algorithms, realistic lighting, materials, textures and other effects are added to the input view to generate a rendered new view, making the rendered view more vivid and realistic, and improving the overall visual experience. After the rendered new view is generated, the rendered new view can be sent to the scene construction component.
[0022] The scene construction component is responsible for receiving the input view, performing style selection (i.e. allowing the target user to select different styles), and obtaining a stylized view so that the input view has a specific visual effect and form of expression. Then, it also receives the rendered new view generated by the rendering component, merges the stylized view with the rendered new view, and obtains the scene view.
[0023] In the scene construction component, style selection allows the target user to select different styles for the input view, so that the final generated view has a specific visual effect and form of expression. The style can be used to stylize the input view, where the style type can include: The realistic style pursues a highly restored real-world visual effect, emphasizes realistic light and shadow effects and fine details, and is suitable for scenes that require high authenticity.
[0024] Cartoon style focuses on simplification and exaggeration, and often uses bright colors and smooth lines.
[0025] Sketching style, using a single color or few tones, imitating the line and shading effects of pencil or charcoal.
[0026] Watercolor style simulates the effect of watercolor painting, with elegant colors and soft edges.
[0027] Cyberpunk style usually has a futuristic, technological and dark tone.
[0028] The retro style creates a nostalgic atmosphere by using faded colors and old textures. It is suitable for historical reproduction, cultural relics display and nostalgic product design.
[0029] The Impressionist style is characterized by an emphasis on changes in light and color, with images often appearing slightly blurred but richly colored.
[0030] The low-polygon style simplifies the geometry of objects by using fewer polygons to create a unique visual effect.
[0031] The abstract style focuses on the free expression of form and color rather than the concrete reproduction of reality. It is suitable for artistic creation, modern art exhibitions and design thinking inspiration.
[0032] The realistic style pursues extremely high image details and lighting processing, making the generated images close to the effect of photographic photos. It is suitable for product display, virtual reality scene reproduction and high-end advertising production.
[0033] Through these style selections, the scene construction component can generate visual effects with specific stylization according to the needs of target users and application scenarios, thereby enhancing the expressiveness and attractiveness of the view.
[0034] The view component is the main interface for the target user to interact with the system. The view component is responsible for displaying the final generated scene view to the target user, thereby providing an intuitive visualization effect to the target user. The view component is also responsible for receiving the target user's interaction requests, such as querying information, performing operations, etc., and passing these requests to the large language model component.
[0035] Exemplarily, a view component is used to receive and process an input view of a target user; a 3D Gaussian uses a Gaussian representation to process 3D data; a 3D Gaussian distribution is used to visualize a 3D Gaussian; a style selection is used to select different styles for visual presentation; a generalizable Gaussian splashing style is used to incorporate stylized selection into Gaussian splashing; a novel view is presented to present a new view based on processed data; a stylized 3D Gaussian is used to apply stylization to a 3D Gaussian representation; and a stylized novel view is used to present a novel view through an applied style.
[0036] The large language model component is responsible for receiving the interaction request sent by the target user through the view component. Then, by connecting with the knowledge graph component, it retrieves and queries the data related to the interaction request from the knowledge graph component, and uses the powerful language understanding and generation capabilities of the large language model component to determine and generate the corresponding response data. At the same time, the view component also displays the corresponding response data to ensure that the target user can obtain the required information feedback in a timely manner.
[0037] The knowledge graph component stores a large amount of data and information, providing data support for the large language model component, enabling it to obtain accurate and comprehensive knowledge and information when processing interaction requests.
[0038] The scenario building component includes "processing input views", and the large language model component includes "storing session data", "generating natural language responses", "processing natural language queries", and "querying data from the knowledge graph". The specific interaction relationships and processes are as follows: First, a preliminary analysis of the input views is processed.
[0039] After receiving the input view (such as a 3D image, scene description, or related parameters) from the target user, the scene construction component first performs preliminary analysis and processing, which may include identifying key elements and user intent in the input view and extracting the information as structured data or natural language description.
[0040] For example, if the target user uploads an image of a wind power generation scene, key information such as wind turbines and environmental conditions can be extracted.
[0041] Second, process natural language queries.
[0042] After the initial analysis, the extracted information can be submitted to a large language model component in the form of natural language for further processing and understanding.
[0043] A large language model component parses the natural language input and identifies the specific needs of the target user (for example, the target user wants to know the optimal operating settings for the current wind turbine).
[0044] Third, query data from the knowledge graph.
[0045] Based on the results of processing natural language queries, the large language model component needs to obtain relevant data from the knowledge graph, which stores a wide range of domain knowledge, including optimal operating parameters for wind farms, equipment maintenance records, environmental impact analysis, etc.
[0046] The large language model component makes requests to retrieve the necessary knowledge data. The knowledge graph interface returns optimization information or action suggestions related to the input view.
[0047] Fourth, generate natural language responses.
[0048] Once the required data is retrieved from the knowledge graph, the large language model component converts the data into natural language descriptions or instructions that are friendly to the target user. These responses can include detailed analysis reports, action suggestions, or answers to the target user's queries.
[0049] For example, the large language model component can arbitrarily generate a recommendation to instruct the target user on how to adjust the turbine settings based on current wind conditions to maximize power generation efficiency.
[0050] Fifth, store session data.
[0051] Throughout the interaction, all inputs, processing steps, and output results are recorded. The large language model component stores this data for future analysis, to optimize model performance, or to provide reference for subsequent queries. Storing session data can also be used to personalize the user experience, allowing the system to remember the target user's preferences and historical queries and provide more accurate recommendations.
[0052] The interaction process between the knowledge graph component and the large language model component is as follows: After receiving the natural language query of the target user, the large language model component first parses the user's intent through the natural language understanding module to identify the entities, relations, and contexts involved in the query.
[0053] The structured information parsed by the large language model component can contain requirements for specific information in the knowledge graph component, such as the need to retrieve the attributes of an entity, find a certain relationship, or obtain rules for a specific field.
[0054] Then, data is obtained from the knowledge graph component, which contains information about entities (such as people, places, objects, etc.) and their attributes and relationships. It can retrieve relevant information from its rich semantic network based on the request of the large language model component.
[0055] The large language model component sends a query request to the knowledge graph component based on the parsed requirements. The request is usually structured, such as a SPARQL query statement, which is used to retrieve specific entity information, attributes or relationships from the knowledge graph component.
[0056] Then, data fusion and natural language generation are performed. The knowledge graph component returns the query result, which is highly structured data, such as the attribute value of an entity, the relationship between two entities, or a set of certain predefined rules.
[0057] After receiving the structured data returned from the knowledge graph component, the large language model component combines this data with the context of the natural language query to generate a natural language response suitable for user understanding. This step is usually completed by the natural language generation module.
[0058] For example, a user enters “What is the best setting for a wind turbine at the current wind speed?”; the large language model component parses the query and identifies the key entities (wind speed, wind turbine) and the type of information required (optimal setting); the large language model component sends a query request to the knowledge graph component to obtain rules or historical data related to wind speed and wind turbine operation; the knowledge graph component returns data about the relationship between wind speed and turbine settings; the large language model component converts this data into a natural language response, such as “Based on the current wind speed, it is recommended to adjust the wind turbine blade angle to 30 degrees to maximize power generation efficiency.” Finally, after completing an interaction, the large language model component can store the query results and generated responses in its memory module so that future related queries can be responded to more quickly and accurately. These memories can further enrich the content of the knowledge graph, in turn improving the overall intelligence of the system.
[0059] The knowledge graph component can also update its internal data structure through feedback from the large language model component, such as adding new entity relationships or updating existing data.
[0060] Therefore, the interaction between the knowledge graph component and the large language model component is bidirectional. The large language model component uses the structured information of the knowledge graph component to enhance its language understanding and generation capabilities, while the knowledge graph component obtains contextual information from the queries of the large language model component to continuously update and optimize its internal data.
[0061] The large language model component includes a large language model. The interaction process between natural language understanding and the large language model includes: Natural language understanding first receives the target user's natural language input and parses out the key intent and information. The parsed information is passed to the large language model as input for further processing. The large language model uses this structured information to generate contextual understanding and build a more comprehensive semantic model. This structured information helps the large language model accurately identify user needs in the subsequent processing process and generate natural language responses that match them.
[0062] From the above, it can be concluded that the rendering of the input view by the rendering component and the fusion of the stylized view and the rendered new view by the scene construction component provide the target user with a visual scene view, which greatly improves the visual experience. Secondly, the connection between the view component and the large language model component enables the target user's interaction request to be responded to in a timely manner. With the help of the rich data in the knowledge graph component, the large language model component can accurately determine the response data corresponding to the interaction request, thereby achieving efficient interactive operations and meeting the diverse needs of users. Therefore, the close collaboration and efficient integration between the components optimize the overall performance of the system and greatly improve the efficiency of interactive operations.
[0063] In one embodiment of the present disclosure, reference Figure 2 , Figure 3 and Figure 4 ,A wind power generation intelligent interactive integration system based on a large language model also includes a robot assistant component; The robot assistant component is connected to the large language model component, and is used to respond to the command of the target user, perform retrieval based on the large language model component, determine the control instruction corresponding to the command based on the retrieval result, and control the external robot to perform the action corresponding to the control instruction according to the control instruction.
[0064] In this embodiment, the target user's command can be voice or text. When the target user issues a command, the robot assistant component will respond immediately.
[0065] First, the robot assistant component performs retrieval based on the large language model component. The large language model component stores a large amount of information and data, which provides support for the retrieval of the robot assistant component. Then, based on the results of the retrieval, the robot assistant component can determine the control instructions corresponding to the target user's command. This process requires analyzing and understanding the retrieved information to accurately convert the target user's command into specific and executable control instructions. Finally, the robot assistant component passes the generated control instructions to the external robot to enable it to perform the corresponding action.
[0066] In the wind power generation scenario, the application scenarios of the robot assistant component can significantly improve the operational efficiency, maintenance quality and the intelligence level of the overall system. The application scenarios of the robot assistant component are described as follows: (1) Routine inspection and maintenance of wind turbines.
[0067] The robotic helper components can be equipped with high-resolution cameras and sensors to regularly inspect the wind turbine’s blades, nacelle and tower structure to detect potential problems such as wear, cracks and corrosion.
[0068] The robot assistant component integrates with the wind sensor and data monitoring module to collect real-time data and transmit the inspection results to the control module. If an abnormality is detected, the robot assistant component can automatically generate a maintenance report and recommend or perform necessary maintenance operations.
[0069] (2) Calibration and debugging of wind sensors Wind sensors need to be calibrated regularly to ensure the accuracy of their measurements. The robotic assistant component can perform automatic calibration and commissioning of wind sensors, reducing the need for human intervention.
[0070] Using a standardized calibration procedure, the robotic assistant component works in conjunction with the control module to adjust the wind sensor settings under different wind speed conditions and verify its accuracy. Once completed, the robotic assistant component generates a calibration report and uploads it to the database.
[0071] (3) Real-time monitoring and data analysis of wind farms The robot assistant component monitors the operating status of the wind farm in real time through the integrated data monitoring module, including wind speed, wind direction, power generation and equipment status.
[0072] The robot assistant component collects various types of data and analyzes them in real time, predicting the performance trends of the equipment through machine learning algorithms and identifying potential failure risks. The robot assistant component can also interface with the knowledge graph component to optimize the operating parameters of the wind farm using industry standards and historical data.
[0073] (4) Rapid response to emergencies When a wind farm encounters extreme weather or sudden equipment failure, the robot assistant component can automatically initiate emergency response procedures to ensure the safe operation of the wind farm.
[0074] The robot assistant component automatically executes the shutdown or restart procedures of the wind turbine based on real-time data and preset emergency plans, and assists on-site personnel in troubleshooting and repairs. The robot assistant component can also report detailed information of the event to the control module to ensure that subsequent response measures are in place.
[0075] (5) Remote control and optimization of power generation equipment Through remote connection, the robot assistant component can control wind power equipment from any location outside the control room, realizing remote operation and system optimization.
[0076] The robot assistant component analyzes the current power generation efficiency based on the real-time data collected, and optimizes the overall output of the wind farm by adjusting the operating parameters of the wind turbine, such as blade angle and generator load. In addition, the robot assistant component can also remotely start, stop or adjust the working mode of the wind power equipment after receiving instructions from the control module.
[0077] (6) Intelligent maintenance and autonomous learning The robot assistant component continuously collects and analyzes the operating data of the wind farm, conducts autonomous learning, and improves the intelligence of maintenance operations.
[0078] The robot assistant component combines historical maintenance data with real-time data and uses machine learning algorithms to optimize maintenance strategies. For example, the robot assistant component can adjust the maintenance cycle based on the frequency of use and failure rate of the equipment, thereby extending the service life of the equipment and reducing maintenance costs.
[0079] (7) Predictive maintenance in wind power generation scenarios By leveraging the predictive capabilities of robotic assistant components, potential equipment failures can be identified in advance, avoiding unplanned downtime.
[0080] The robot assistant component continuously analyzes the operating status of the equipment through the data monitoring module, combines industry standards and historical fault data in the knowledge graph, predicts possible equipment failures, and arranges maintenance work in advance to ensure the continuity of wind power generation.
[0081] By deploying robotic assistant components in wind power generation scenarios, the operating efficiency, equipment maintenance quality and overall management level of wind farms can be significantly improved, while also reducing the risks and costs of manual operations.
[0082] Exemplarily, the target user's command is voice. The robot assistant component converts the target user's voice into text for processing through voice recognition; then, natural language understanding is used to understand and interpret the target user's voice to achieve dialogue management, that is, to manage the interaction and dialogue process with the target user; then, according to the target user's voice and the large language model component, relevant information is queried and retrieved for knowledge retrieval, so as to make a decision based on the retrieval results and the target user's voice; finally, the external physical robot is controlled based on the decision and the target user's voice.
[0083] Among them, external physical robots refer to physical robots with autonomous mobility, operation and perception capabilities, which can perform various tasks in the physical world, including inspection, maintenance, monitoring and troubleshooting. Specifically, the application of external physical robots in wind power generation scenarios includes: Inspection robots, equipped with cameras, sensors and other detection equipment, are able to climb the tower, blades or nacelle of a wind turbine and perform physical inspections to detect damage or wear on the structure.
[0084] Maintenance robots, capable of carrying tools and parts, can perform complex repair tasks on site, such as replacing parts of a turbine, repairing cracks or adjusting equipment settings.
[0085] Inspection robots, which usually have automatic navigation and obstacle avoidance functions, can patrol wind farms, monitor the operating status of equipment in real time, and identify and report abnormal situations.
[0086] Emergency response robots can respond quickly to emergencies (such as fires or extreme weather) at wind farms and perform emergency shutdowns, cooling or other protective measures to ensure the safety of the wind farm.
[0087] Teleoperated robots, where operators can remotely control the robot to perform tasks that are not suitable or possible to be directly operated by humans, such as maintenance work at high altitudes, inclement weather or dangerous environments.
[0088] Robotic assistant components can significantly reduce manual workload, lower maintenance costs, and improve the safety and reliability of wind power facilities.
[0089] The large language model component integrates the backend, large language model, knowledge graph interface, memory, dialogue manager, natural language generation, reasoning engine, data processing and training modules. Among them, the backend is used to manage the backend operations and communication between components. The large language model is the core language model of the large language model component, which can be an instance of ChatGPT-4, Llama 3, etc. The knowledge graph interface is connected to the knowledge graph for information retrieval and integration. The memory is responsible for managing the storage and retrieval of information within the system. The dialogue manager manages the dialogue with the target user to ensure coherent and contextual responses. Natural language generation generates natural language responses based on the target user's commands and system data. The reasoning engine uses the large language model to perform reasoning tasks. It is used by the system through data processing. The training module is responsible for managing the training of the large language model, integrating new data and updates. The integration of various modules in the large language model component provides convenience for interaction with other components.
[0090] From the above, we can conclude that by introducing the robot assistant component, the execution capability of the system is greatly enhanced. The commands of the target user can be directly converted into the actual actions of the external robot, which improves work efficiency. By connecting with the large language model component and accurately retrieving it, the accuracy and rationality of the control instructions are ensured, and the risk of erroneous operation is reduced. At the same time, it also provides a more convenient operation experience for the target user. No complicated settings are required. The control of the external robot can be achieved by simply issuing commands, saving time and energy.
[0091] In one embodiment of the present disclosure, reference Figure 2 and Figure 4 , the robot assistant component is connected with the knowledge graph component; The robot assistant component is also used to collect environmental data and send the environmental data to the knowledge graph component; The knowledge graph component is used to update the knowledge base based on environmental data.
[0092] In this embodiment, the robot assistant component can not only determine the control instructions based on the information of the large language model component, but also has the function of collecting environmental data. The robot assistant component can collect various real-time data of the environment through various sensors. Then, the collected environmental data is sent to the knowledge graph component. After receiving the environmental data, the knowledge graph component will analyze and process it, integrate the new data into the existing knowledge base, and update the knowledge base.
[0093] In wind power generation scenarios, the robot assistant component can collect various real-time data about the environment through a variety of sensors, which helps optimize the operation of wind farms, perform predictive maintenance, and improve safety and efficiency. The types of environmental data and corresponding sensors include: (1) Wind speed and direction Data type: wind speed, wind direction; Sensors: anemometer, wind vane; Purpose: Used to monitor the operating conditions of wind turbines in real time and help adjust the blade angle and speed of the turbines to optimize power generation efficiency.
[0094] (2) Temperature Data type: ambient temperature, equipment surface temperature; Sensor: temperature sensor; Purpose: Monitor ambient and device temperatures to prevent equipment from overheating and take protective measures in extreme weather conditions.
[0095] (3) Humidity Data type: air humidity; Sensor: humidity sensor; Application: To monitor humidity in the air and prevent equipment corrosion or material deterioration caused by high humidity.
[0096] (4) Vibration Data type: vibration frequency, amplitude; Sensors: accelerometer, vibration sensor; Purpose: Detect abnormal vibrations in wind turbines and other mechanical components to identify potential mechanical failures in advance.
[0097] (5) Sound Data type: environmental noise, equipment operation noise; Sensors: microphones, acoustic sensors; Purpose: To monitor the sound of equipment during operation and detect abnormal noise, thereby detecting faults or damage in advance.
[0098] (6) Light intensity Data type: light intensity; Sensor: light sensor; Purpose: Monitor light conditions at a wind farm, which is especially important when maintenance or visual inspections are required.
[0099] (7) Pressure Data type: pneumatic pressure, hydraulic pressure; Sensor: pressure sensor; Purpose: To monitor pressure changes inside the equipment or in the environment to ensure that the system operates within a safe range.
[0100] (8) Distance and location Data type: distance to the object, position coordinates; Sensors: laser rangefinder, ultrasonic sensor, GPS module; Purpose: Measure the tilt or deformation of wind turbines, monitor the real-time position and navigation of external physical robots.
[0101] (9) Rainfall Data type: precipitation; Sensor: Rain gauge; Purpose: Monitoring rainfall to help predict potential impacts on wind power systems, such as equipment waterproofing requirements.
[0102] (10) Electromagnetic fields Data type: electromagnetic field strength; Sensor: electromagnetic field sensor; Purpose: To monitor possible electromagnetic interference in wind farms to ensure normal operation of equipment and protect the safety of workers.
[0103] (11) Inclination Data type: tilt angle of the device; Sensor: Tilt sensor; Purpose: Monitor the tilt angle of wind turbine towers to prevent collapse or damage due to structural instability.
[0104] Exemplarily, data is collected from various sensors and preprocessed, and the collected data is used for further analysis and use. Then, data integration is performed to integrate the processed data into the system. Finally, a knowledge base is generated to store knowledge specifically for the robot assistant component. At the same time, continuous learning is used to ensure that the robot assistant component continues to learn and improve, so that analytical feedback is performed to improve decision-making and system performance, and the model of the robot assistant component is updated according to new data and feedback, so that the performance of the robot assistant component is continuously optimized.
[0105] From the above, we can conclude that the robot assistant component collects environmental data and transmits it to the knowledge graph component, so that the knowledge base can be updated in real time, ensuring the timeliness and accuracy of information, which is conducive to the system to adapt to environmental changes more accurately and make reasonable decisions. At the same time, it also provides a strong driving force for the intelligent development of the system, allowing the robot assistant component to continuously learn and evolve, better serve users and meet various complex needs.
[0106] In one embodiment of the present disclosure, reference Figure 2 , Figure 5 , Figure 6 and Figure 7 ,A wind power generation intelligent interactive integration system based on a large language model also includes a wind power generation scenario component; The wind power scenario component is connected to the large language model component; The view component is also used to receive a first request from a target user; The large language model component is also used to determine wind power generation scenario data from the wind power generation scenario component in response to the first request; The view component is also used to display wind power generation scenario data.
[0107] In this embodiment, the first request is "request for wind power generation scenario data". When the target user sends the first request through the view component, the large language model component will respond to it. The large language model component will obtain relevant data from the wind power generation scenario component connected to it, which may include real-time wind power generation power, wind turbine operating status, wind speed and direction, etc. The view component will then receive the wind power generation scenario data from the large language model component and display it to the user, so that the user can intuitively understand the relevant situation of wind power generation.
[0108] Exemplarily, the target user sends a first request to the view component, and then the large language model component responds to the first request. The large language model component requests data from the wind power generation scenario component, and the data monitoring module collects data and returns the collected data to the wind power generation scenario component. The wind power generation scenario component can also request maintenance data from the maintenance module, and the maintenance module returns the maintenance data to the wind power generation scenario component. The wind power generation scenario component returns the scenario data to the large language model component, and at the same time, the large language model component returns the wind power generation scenario data to the view component.
[0109] Exemplarily, the system can be in various states. When idle, it indicates that the system is in a waiting state and is ready to receive target user input or instructions; when the target user input has been received, the system has received the target user's input and is processing the request; start scene construction, the system starts to build the scene according to the target user input; create 3D Gaussian, the system creates a 3D Gaussian object according to the input view; apply style selection, the system applies the selected style to the scene; Gaussian sputtering, the system performs sputtering on the 3D Gaussian; rendering, the system generates novel views and stylized novel views; large language model processing, the system interacts with the large language model to process data and generate responses; reasoning execution, the system uses the large language model to perform reasoning tasks; knowledge graph query, the system queries the knowledge graph to obtain relevant information; memory management, the system manages memory operations, such as storing and retrieving information; data processing, the system processes data input for further analysis; robot assistant operation, the system performs tasks related to the robot assistant; the system analyzes feedback to improve decision-making; the system updates the model based on new data and feedback; the system monitors and manages wind power scenes; the system performs maintenance activities on wind power generation systems; the process ends and the system returns to the idle state.
[0110] From the above, it can be concluded that the wind power generation scenario component provides a convenient way for target users to obtain wind power generation scenario data. Target users only need to send a request through the view component to quickly obtain the required information, which improves the user experience and allows users to grasp the situation of wind power generation in real time and intuitively, which helps to find problems and make decisions in time, and improve power generation efficiency and safety.
[0111] In one embodiment of the present disclosure, reference Figure 2 and Figure 4 ,The wind power generation scenario components include wind turbines, wind sensors, control modules, data monitoring modules, and maintenance modules; The wind turbine is connected to a wind sensor, the wind sensor is connected to a control module, the control module is connected to a data monitoring module and a maintenance module respectively; the control module, the data monitoring module, and the maintenance module are all connected to a large language model component; The wind sensor is used to collect wind condition data and send the wind condition data to the control module; The data monitoring module is used to monitor the operation data, and the maintenance module is used to manage the maintenance data; Among them, wind power generation scenario data includes operation data or maintenance data.
[0112] In this embodiment, the wind sensor is connected to the wind turbine, and can collect wind data in real time, such as wind speed, wind direction, etc., and send the collected data to the control module. The control module is responsible for receiving the data from the wind sensor, so as to regulate the operation of the wind turbine. At the same time, the control module is respectively connected to the data monitoring module and the maintenance module to realize information interaction. Among them, the data monitoring module closely monitors the operating data of the system, including power generation, equipment temperature, etc., to ensure the normal operation of the system. The maintenance module is responsible for managing maintenance data, such as equipment maintenance records, maintenance plans, etc. By connecting with the large language model component, the relevant data is passed to the large language model component to provide data support for the intelligent decision-making and interaction of the system.
[0113] Exemplarily, the wind sensor is responsible for monitoring wind conditions and providing data to the control module. The power converter can convert wind energy into usable electrical energy, and then connect the generated electrical energy to the grid to achieve power supply. The control module is responsible for managing the control of the wind power generation scenario components, thereby monitoring the data of the wind power generation scenario components and managing the maintenance activities of the wind power generation scenario components. The wind power generation scenario components are integrated into the entire system by interacting with the large language model component to perform control and monitoring tasks.
[0114] The specific process of the wind power generation scenario component performing control and monitoring tasks by interacting with the large language model component is as follows: First, the dynamic adjustment of wind turbines.
[0115] The wind speed sensor and wind direction sensor in the wind power generation scenario component monitor the current wind speed and wind direction data in real time.
[0116] The wind power scenario component transmits this data to the large language model component, which analyzes the optimal wind turbine blade angle and speed settings through its built-in reasoning engine and knowledge graph interface.
[0117] The large language model component generates a set of optimized control commands that instruct the wind turbine to adjust the blade angle and speed to maximize power generation efficiency.
[0118] The adjusted power generation efficiency and equipment status data are transmitted back to the large language model component for further optimization and adjustment of future control strategies.
[0119] Second, detection and response to abnormal vibrations.
[0120] The vibration sensors of the wind power generation scene component monitor the vibration frequency and amplitude of the turbine and other mechanical parts in real time. The data is transmitted to the large language model component, which uses the pre-trained model and knowledge graph analysis to determine whether the current vibration is abnormal.
[0121] If abnormal vibrations are detected, the large language model component generates a detailed alarm message including possible fault causes and recommended maintenance actions.
[0122] The alarm information notifies the on-site operator through the view component and is simultaneously passed to the robot assistant component, triggering it to perform automatic inspection or maintenance tasks.
[0123] The robotic assistant component performs specific maintenance actions, such as inspecting mechanical parts of a wind turbine or adjusting equipment settings, based on the analysis results of the large language model component.
[0124] Third, prediction of the impact of weather changes on power generation.
[0125] The wind power generation scenario component collects real-time weather data from environmental sensors (such as temperature, humidity, light intensity, etc.). The weather data is transmitted to the large language model component, which is analyzed through knowledge graphs and weather models to predict the impact of future weather changes on wind power generation efficiency.
[0126] The large language model component generates adjustment strategies, including possible generation power adjustments and equipment protection measures (such as suspending generation under extreme weather conditions).
[0127] The control module executes the adjustment strategy and performs necessary equipment inspections and preventive maintenance via the robotic assistant component.
[0128] The actual impact of weather changes is continuously monitored, and feedback data is passed back to the large language model component for optimization of future strategies.
[0129] Fourth, remote monitoring and maintenance.
[0130] The sensors and data monitoring modules in the wind power scenario component continuously provide data streams to the large language model component, including equipment status, environmental conditions, and power generation performance. The large language model component integrates this data and analyzes the overall operating status of the current power plant through machine learning algorithms to identify potential problems.
[0131] If a problem is detected, the large language model component generates a detailed diagnostic report and provides guidance to the remote operation and maintenance team through the view component, explaining the specific measures to be taken. At the same time, the large language model component can also generate automated maintenance instructions and perform necessary on-site inspections and maintenance operations through the robot assistant component.
[0132] The execution results are fed back to the large language model component in real time to further optimize the model's diagnostic and predictive capabilities.
[0133] Through these interactive processes, the connection between the wind power generation scenario and the entire system is enhanced, so that it not only exists as an independent component, but is tightly integrated into a large-scale language model-driven intelligent control and monitoring system, which can significantly improve the automation level and operational efficiency of wind farms.
[0134] From the above, it can be concluded that the wind sensor collects wind data and transmits it to the control module, so that the wind turbine can optimize its operation according to the real-time wind conditions and improve power generation efficiency. The data monitoring module monitors the operation data, which helps to detect potential faults in time, ensure the stable operation of the system, and reduce downtime losses. The maintenance module manages maintenance data, which can reasonably arrange equipment maintenance plans, extend equipment service life, and reduce maintenance costs. Wind power generation scenario data covers operation and maintenance data, providing comprehensive information for large language model components, helping to realize intelligent power generation management and decision-making, and improving the reliability of the entire wind power generation system.
[0135] In one embodiment of the present disclosure, reference Figure 2 , Figure 5 , Figure 6 and Figure 7 , a wind power generation intelligent interactive integration system based on a large language model also includes: The view component is also used to receive a second request from the target user; The large language model component is further used to determine the target scenario based on the scenario building component in response to the second request; The view component is also used to display the target scene.
[0136] In this embodiment, the second request is "request scene visualization". When the target user issues the second request through the view component, the large language model component will respond to the second request. At the same time, the large language model component will work with the scenario construction component to determine the target scene. The scenario construction component will use its functions and data to generate or select a target scene that meets the requirements based on the needs of the target user, the system settings, and related conditions and rules. After that, the view component will receive information about the target scene and present it to the user in a visual form. The target user can intuitively see the requested scene, so as to better understand and interact with the system.
[0137] Exemplarily, the target user sends a second request to the view component, the view component calls the scene visualization to visualize the scene, and the scene visualization sends a request scene data command to the scene construction component. The scene construction component sends a "get rendered view" command to the rendering view component, and then the rendering view component feeds back a "return rendered view" to the scene construction component, after which the scene construction component sends a "get stylized view" to the stylized view, and the stylized view is returned to the scene construction component, and finally the scene construction component returns the scene data to the scene visualization, and finally displays the scene through the view component.
[0138] It can be concluded from the above that the target user can obtain the required specific scene display by issuing the second request, which can improve the target user's ability to understand and grasp complex information, and the intuitive target scene presentation makes it easier for the target user to understand the operating status of the system. Therefore, this embodiment not only helps to improve the accuracy and efficiency of decision-making, but also enhances the interactivity and user experience of the system.
[0139] In one embodiment of the present disclosure, reference Figure 2 , Figure 5 , Figure 6 and Figure 7 , a wind power generation intelligent interactive integration system based on a large language model also includes: The view component is also used to receive a third request from the target user; The large language model component is further used to update the scene data based on the scene component in response to a third request; The view component is also used to display updated scene data.
[0140] In this embodiment, the third request is "request scene reload". When the target user issues the third request through the view component, the large language model component will respond immediately. The large language model component will work in conjunction with the scene component. The scene component is responsible for storing and managing scene-related data. The large language model component will send instructions to the scene component to request an update of the scene data. This may include obtaining the latest real-time data, recalculating certain parameters, loading new configuration information, or correcting previous data errors. After the update is completed, the view component will display the updated scene data in a timely manner to ensure that users can see the latest and most accurate scene information.
[0141] Exemplarily, the target user sends a third request to the view component, and the scene is reloaded by starting reloading. Then, through the scene reloading, a "request scene data" command is sent to the scene construction component, the scene construction component sends a "get rendering view" command to the rendering view component, and then the rendering view component feeds back "return rendering view" to the scene construction component, and then the scene construction component sends "get stylized view" to the stylized view, and the stylized view is returned to the scene construction component, and finally the scene construction component returns the scene data to the scene reloading, and finally the scene data is displayed through the view component.
[0142] From the above, we can conclude that the view component ensures that the target users can always obtain the latest scene data and maintain the accuracy and timeliness of information in an environment where information changes rapidly. For complex scenes, it can help users make timely and accurate decisions. At the same time, it not only improves the target user experience, allowing the target users to easily refresh the scene without complex operations and keep abreast of the latest developments, but also enhances the flexibility and adaptability of the system. Whether it is data update or scene configuration adjustment, it can respond quickly to meet the ever-changing needs of users.
[0143] In one embodiment of the present disclosure, reference Figure 2 and Figure 4 The rendering component is specifically used to render the input view based on the 3D Gaussian sputtering algorithm to obtain a rendered new view.
[0144] In this embodiment, the steps of the 3D Gaussian sputtering algorithm include: preparing raw data for further processing through a preprocessing module; extracting relevant features from the preprocessed data through a feature extraction module; normalizing the data to ensure consistency through a normalization module; dividing the spatial data into manageable parts through a spatial subdivision module; defining the Gaussian kernel used in sputtering through a Gaussian kernel definition; calculating the weights of the sputtering process through a weight calculation module; applying the Splatting technique to 3D data through a splatting module; integrating the scattered data to generate the final output through a fusion module; storing the final processed data for rendering; and finally, outputting the processed data rendering view through a rendering view. The 3D Gaussian sputtering algorithm processes data for rendering and visualization tasks, which are then visualized by the scene building component.
[0145] From the above, we can conclude that the 3D Gaussian sputtering algorithm can bring a more immersive experience to the target users. At the same time, it improves the efficiency of rendering, and reduces the time and computing resources required for rendering while ensuring high quality. Therefore, it helps the system respond quickly, allowing users to obtain exquisite rendering results in time and better interact and analyze.
[0146] In one embodiment of the present disclosure, reference Figure 2,The knowledge graph component includes a knowledge base that includes external data, historical data and industry data; The large language model component is connected to the knowledge graph component through the knowledge graph interface.
[0147] In this embodiment, the knowledge base refers to the central repository of structured knowledge used by the system. External data refers to interfacing with external data sources to obtain more information. Historical data refers to storing and managing historical data for reference and analysis. Industry standards refer to industry standards and guidelines related to system operations. The output of the knowledge graph refers to providing the output and results of knowledge graph queries. The large language model component is connected to the knowledge graph component through the knowledge graph interface to perform knowledge graph operations, and its operation process includes: knowledge base query, external data integration, historical data access, and industry standard reference. Among them, knowledge base query refers to the system querying the knowledge base to obtain structured information. External data integration refers to the system integrating external data sources. Historical data access refers to the system accessing historical data for reference. Industry standard reference means that the system refers to industry standards.
[0148] From the above, it can be concluded that the rich knowledge base provides the system with comprehensive and structured knowledge, covering external data, historical data and industry data, and enhances the system's information storage and analysis capabilities. The large language model component is connected to it through an interface to implement operations such as knowledge base query and external data integration, which can obtain more extensive and accurate information and improve the quality and accuracy of answers. Historical data access facilitates the analysis of trends and rules and provides a reference for decision-making. External data integration enriches the data source, and industry standard references ensure that the system operates in compliance with specifications. Therefore, this embodiment improves the intelligence level and practicality of the system and is suitable for a variety of scenarios.
[0149] In one embodiment of the present disclosure, reference Figure 2 and Figure 4 , the view component is also used to display and render new views.
[0150] In this embodiment, the view component not only displays the stylized new view but also displays the rendered new view. Rendering the new view is to generate a new view of the scene from a different angle or viewing angle, which can be obtained by a 3D Gaussian sputtering algorithm.
[0151] From the above, it can be concluded that the scene views displayed by the view components are rich and diverse, which provides good support for scene visualization and greatly enhances the user experience.
[0152] The present invention relates to an integrated system that combines scene construction, large language model interaction and robot assistance. The system applies 3D Gaussian sputtering algorithm, large language model, retrieval enhancement generation and knowledge graph to create an interactive intelligent environment. The system enhances user experience by providing real-time data visualization, responsive robot interaction and seamless integration of various data sources. At the same time, the system also provides efficient data processing capabilities.
[0153] The above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them. Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present disclosure.
Claims
1. A wind power generation intelligent interactive integration system based on a large language model, characterized in that: include: View component, scene building component, rendering component, large language model component and knowledge graph component; The rendering component is connected to the scene construction component and is used to render the input view of the target user to obtain a rendered new view; The scene construction component is connected to the view component and is used to stylize the input view to obtain a stylized view, and merge the stylized view with the rendered new view to obtain a scene view; The view component is used to display the scene view; The view component is connected to the large language model component and is used to receive an interaction request from a target user; The large language model component is connected to the knowledge graph component, and is used to respond to the interaction request and determine the response data corresponding to the interaction request based on the data in the knowledge graph component; The view component is also used to display the response data.
2. The wind power generation intelligent interactive integration system based on a large language model as claimed in claim 1, characterized in that: Also included is a robotic assistant component; The robot assistant component is connected to the large language model component, and is used to respond to the command of the target user, perform retrieval based on the large language model component, determine the control instruction corresponding to the command based on the retrieval result, and control the external robot to perform the action corresponding to the control instruction according to the control instruction.
3. The wind power generation intelligent interactive integrated system based on a large language model as claimed in claim 2, characterized in that: The robot assistant component is connected to the knowledge graph component; The robot assistant component is also used to collect environmental data and send the environmental data to the knowledge graph component; The knowledge graph component is used to update the knowledge base according to the environmental data.
4. The wind power generation intelligent interactive integrated system based on a large language model according to any one of claims 1 to 3, characterized in that: It also includes wind power generation scenario components; The wind power generation scenario component is connected to the large language model component; The view component is also used to receive a first request from a target user; The large language model component is further used to determine wind power generation scenario data from the wind power generation scenario component in response to the first request; The view component is also used to display the wind power generation scene data.
5. The wind power generation intelligent interactive integrated system based on a large language model as claimed in claim 4, characterized in that: The wind power generation scenario components include a wind turbine, a wind sensor, a control module, a data monitoring module and a maintenance module; The wind turbine is connected to the wind sensor, the wind sensor is connected to the control module, the control module is connected to the data monitoring module and the maintenance module respectively; the control module, the data monitoring module and the maintenance module are all connected to the large language model component; The wind sensor is used to collect wind condition data and send the wind condition data to the control module; The data monitoring module is used to monitor the operation data, and the maintenance module is used to manage the maintenance data; Wherein, the wind power generation scenario data includes the operation data or the maintenance data.
6. The wind power generation intelligent interactive integrated system based on a large language model as claimed in claim 1, characterized in that: Also includes: The view component is also used to receive a second request from a target user; The large language model component is further used to determine a target scenario based on the scenario building component in response to the second request; The view component is also used to display the target scene.
7. The wind power generation intelligent interactive integrated system based on a large language model as claimed in claim 1, characterized in that: Also includes: The view component is also used to receive a third request from a target user; The large language model component is further used to update the scene data based on the scene component in response to the third request; The view component is also used to display the updated scene data.
8. The wind power generation intelligent interactive integration system based on a large language model as claimed in claim 1, characterized in that: The rendering component is specifically used to render the input view based on a 3D Gaussian sputtering algorithm to obtain a rendered new view.
9. The wind power generation intelligent interactive integrated system based on a large language model as claimed in claim 1, characterized in that: The knowledge graph component includes a knowledge base, which includes external data, historical data and industry data; The large language model component is connected to the knowledge graph component via a knowledge graph interface.
10. The wind power generation intelligent interactive integrated system based on a large language model as claimed in claim 1, characterized in that: The view component is also used to display the rendered new view.
Citation Information
Patent Citations
Power grid risk early warning and decision-making system, method and equipment based on knowledge graph
CN116523294A
Wind turbine generator operation and maintenance management method based on knowledge graph
CN116775895A
Natural language interaction software framework based on knowledge graph and construction method thereof
CN116860985A
Intelligent question and answer interaction system based on semantic analysis engine
CN117056479A
Transformer substation knowledge graph construction and optimization method based on multi-view learning
CN117251582A
Cited By
Large language model retrieval enhancement method and device, electronic equipment and storage medium
CN119829744A
