Pumped storage power station 5G + BIM intelligent interaction management system

Abstract

The invention provides a pumped storage power station 5G + BIM intelligent interaction management system, and the system comprises a sensing layer which is used for collecting earthwork vehicle transportation data, dam rolling data, underground cavern construction data and monitoring data, and achieves the dynamic data updating of a BIM model through a 5G camera, a sensor and construction data; the network layer transmits dynamic data of the BIM model through a 5G private network, and allocates communication resources according to service requirements by adopting a network slicing strategy; the data layer is used for fusing dynamic data of the BIM model, including BIM geometric data, sensor real-time parameters and construction data, and constructing a dynamic digital twinborn database; and the application layer performs construction state early warning and construction scheduling based on design parameters and sensor parameters and a BIM model, generates a decision instruction through a deep learning algorithm, and realizes data deep fusion, network resource optimization and emergency capability improvement by fusing 5G high-reliability communication, BIM dynamic modeling and AI intelligent decision.

Description

Technical Field

[0001] This invention relates to the field of intelligent interactive data management, specifically to a 5G+BIM intelligent interactive management system for pumped storage power stations. Background Technology

[0002] Pumped storage power stations, as important peak-shaving and valley-filling facilities in the power system, play a crucial role in ensuring grid stability and enhancing the absorption capacity of renewable energy. However, the construction of pumped storage power stations is time-consuming, involves complex procedures, and requires numerous pieces of equipment. Existing pumped storage power station management systems suffer from significant deficiencies in data integration, real-time decision-making, and network adaptability. Furthermore, pumped storage projects are typically built in mountainous areas with significant elevation differences and complex terrain, making public grid signals susceptible to obstruction and reflection, leading to unstable transmission.

[0003] On the other hand, the aquatic environment of pumped storage power stations may lead to increased signal propagation loss, while network slicing can allocate communication resources according to service needs. If data collected by underwater equipment in pumped storage projects is directly transmitted to the cloud, the transmission efficiency may be low due to signal attenuation caused by the aquatic environment.

[0004] Therefore, there is an urgent need for an integrated management system that combines 5G high-reliability communication, BIM dynamic modeling, and AI intelligent decision-making to achieve stable signal transmission, deep data fusion, intelligent proactive operation and maintenance, optimized network resources, and improved emergency response capabilities. Summary of the Invention

[0005] To address the technical problems mentioned above, this invention provides a 5G+BIM intelligent interactive management system for pumped storage power stations, characterized by comprising a perception layer, a network layer, a data layer, and an application layer. The perception layer is used to collect equipment implementation parameters, including real-time parameters of water turbines, generators, water pipelines and reservoir water levels, and to realize dynamic data updates of the BIM model through 5G cameras, sensors and underwater inspection equipment. The network layer transmits dynamic data of the BIM model through the 5G private network and uses a network slicing strategy to allocate communication resources according to business needs. The data layer is used to integrate dynamic data from the BIM model, including BIM geometric data, real-time equipment parameters, and historical operation and maintenance data, to build a dynamic digital twin database. The application layer performs predictive maintenance, intelligent scheduling, and risk warning based on equipment parameters and BIM models, and generates decision instructions through deep learning algorithms; It also includes an edge-cloud collaborative computing module, which is used to dynamically allocate computing resources according to the real-time requirements of the task, through distributed training and inference of LSTM neural network model and reinforcement learning-based scheduling model. Edge computing nodes: deployed at 5G base stations, field servers or terminals to perform tasks with high real-time requirements; Edge computing nodes are divided into multiple levels based on their assessed computing power, with different levels corresponding to different computing tasks; Cloud computing center: performs complex computing tasks and feeds the results back to edge nodes via a 5G private network; The cloud computing center provides computing tasks of a corresponding scale based on the edge computing nodes; The specific mechanism of edge-cloud collaboration is as follows: Edge nodes preprocess data and upload key information to the cloud; The AI ​​model is trained in the cloud and then distributed to edge nodes to update the local model.

[0006] In the preferred embodiment, the real-time parameters of the equipment are mapped to the BIM model in real time through a 5G private network to form a dynamic digital twin; By connecting the BIM API with a dynamic digital twin database, real-time parameter-driven status updates of the BIM model can be achieved.

[0007] In the preferred embodiment, the real-time parameters of the turbine include bearing vibration, guide vane opening, head height, and flow rate; A turbine fault prediction model is built. Real-time parameters of the turbine are input, an LSTM neural network model is used to predict the remaining life of the bearing, an anomaly detection algorithm is used to identify abnormal vibration modes, and potential faults and recommended maintenance solutions are output 7 days in advance.

[0008] In the preferred embodiment, underwater inspection equipment equipped with a 5G camera is used to capture videos of the inner wall of the pipeline. The YOLOv8+U-Net joint model is used to detect cracks and cavitation areas and quantify the damage area. The detection results are automatically labeled to the BIM model to generate repair work orders.

[0009] In the preferred scheme, a reinforcement learning-based scheduling model is built to optimize the pumping / power generation switching strategy based on the grid load demand and reservoir water level. The input parameters of the reinforcement learning-based scheduling model include electricity price, water level and unit efficiency curve. The optimal scheduling command is dynamically generated and sent to the unit control module through the 5G network.

[0010] The preferred solution also includes a project progress prediction module, which includes project progress prediction and work order management functions; The project progress forecasting function specifically includes: dynamically adjusting the project schedule based on the overall construction progress, predicting construction steps in advance, preparing materials, personnel and equipment, and generating construction priority work orders; The work order management function specifically includes: automatically dispatching work orders to the responsible person's mobile terminal and using AR to remotely guide installation and maintenance procedures.

[0011] The preferred solution also includes an emergency command module, which simulates extreme conditions based on a BIM model and uses AI to predict the impact range of a dam failure and generate evacuation routes. When the backbone network is interrupted, 5G D2D devices are used to directly connect and build a temporary communication network to ensure the transmission of command instructions.

[0012] In the preferred embodiment, network slicing includes control command slices, video surveillance slices, and BIM data slices; The control command slice carries unit start / stop commands and valve control commands; Video surveillance slices carry video from drone inspections or images or video data collected by underwater inspection equipment; BIM data slices carry dynamic data of the BIM model and AR / VR interactive data.

[0013] In the preferred scheme, the hardware configuration and tasks of the edge computing nodes are obtained, the edge computing nodes are divided into three levels, different levels are configured with different amounts of BIM models, and the tasks are set to the highest priority to ensure real-time response. Different scales of BIM models refer to lightweight BIM models and lightweight local AI models that include only modules related to the work task and parameters related to the work task and match the hardware configuration of the current node. Lightweight BIM models refer to BIM models that minimize redundant data and have lower precision around the work task. Lightweight local AI models refer to models that are used only for data preprocessing related to work tasks or for executing tasks issued by cloud computing centers without iterative training.

[0014] In the preferred embodiment, the distributed training and inference process of the LSTM neural network model includes: The real-time acquired turbine bearing vibration signals are preprocessed to extract key features; a lightweight LSTM model is used for real-time inference to predict the bearing's remaining life RUL(t). The inference formula is as follows: ; in, For LSTM hidden states, For input features, For LSTM inference functions; Historical data from several edge nodes are aggregated, and a federated learning framework is used for global model training. Edge nodes upload model gradients, and the cloud aggregates and updates the global model parameters. ; in, This represents the global model parameters after aggregation in the cloud. These are the initial parameters for the local model. Where is the learning rate, and N is the number of edge nodes. The model gradient calculated for the i-th edge node based on local data; The cloud will update the global model parameters The model is deployed to edge nodes to replace the local model and optimize inference accuracy.

[0015] The 5G+BIM intelligent interactive management system for pumped storage power stations provided by this invention has the following advantages compared with existing technologies: (1) The 5G+BIM intelligent interactive management system of the present invention can make full use of the real-time data of the 5G system, greatly improve the carrying capacity of the BIM system, and enable more design, construction, management, conflict resolution, project progress prediction and other data to be loaded into the BIM system in real time, so as to realize the replacement of manual management with system management and greatly improve the efficiency and accuracy of design, construction and management work. The 5G system can also support the BIM system to carry out collaborative work at a higher granularity.

[0016] (2) While the capabilities of BIM systems are being enhanced, new technical challenges are also emerging. Specifically, terminal devices struggle to respond to BIM system management in real time, and the massive amounts of data pose computational challenges to large-scale BIM systems and related AI systems, leading to an almost unlimited demand for computing power. This invention employs an edge-cloud collaborative computing module, fully utilizing the computing resources of edge nodes and cloud computing centers. Through edge node data preprocessing, grading, and lightweighting, edge nodes can provide a more granular BIM system and a more intelligent interactive experience while ensuring the completion of their tasks.

[0017] (3) Use 5G D2D technology to build a temporary communication network to ensure the transmission of command instructions and ensure rapid response and effective handling in emergency situations.

[0018] (4) By using 5G network slicing technology, different types of service data are allocated to different network slices to ensure the efficiency and reliability of data transmission.

[0019] (5) The edge-cloud collaborative computing module achieves efficient data processing and model updates by rationally allocating computing resources between edge computing nodes and cloud computing centers. This not only improves the real-time performance of the system, but also reduces the dependence on a single computing node and ensures the stable operation of complex computing tasks.

[0020] (6) By means of 5G private network coverage, network slicing resource optimization, edge-cloud collaboration to reduce transmission volume, and D2D technology for emergency blind spot filling, the problems of weak signal coverage caused by terrain difference and transmission loss caused by water environment in pumped storage projects have been solved in a targeted manner, which has significantly improved signal transmission efficiency and stability. Attached Figure Description

[0021] Figure 1 This is the system architecture diagram of the present invention.

[0022] Figure 2 This is a flowchart of the edge-cloud collaborative computing process of the present invention. Detailed Implementation

[0023] like Figures 1-2 As shown, a 5G+BIM intelligent interactive management system for pumped storage power stations includes a perception layer, a network layer, a data layer, and an application layer. The perception layer is used to collect equipment implementation parameters, including real-time parameters of water turbines, generators, water pipelines and reservoir water levels, and to realize dynamic data updates of the BIM model through 5G cameras, sensors and underwater inspection equipment. The network layer transmits dynamic data of the BIM model through the 5G private network and uses a network slicing strategy to allocate communication resources according to business needs. The data layer is used to integrate dynamic data from the BIM model, including BIM geometric data, real-time equipment parameters, and historical operation and maintenance data, to build a dynamic digital twin database. The application layer performs predictive maintenance, intelligent scheduling, and risk warning based on equipment parameters and BIM models, and generates decision instructions through deep learning algorithms; It also includes an edge-cloud collaborative computing module, which is used to dynamically allocate computing resources according to the real-time requirements of the task, and supports distributed training and inference of LSTM neural network models and reinforcement learning-based scheduling models. Edge computing nodes: deployed at 5G base stations, field servers or terminals to perform tasks with high real-time requirements; Edge computing nodes are divided into multiple levels based on their assessed computing power, with different levels corresponding to different computing tasks; Cloud computing center: performs complex computing tasks and feeds the results back to edge nodes via a 5G private network; The cloud computing center provides computing tasks of a corresponding scale based on the edge computing nodes; The specific mechanism of edge-cloud collaboration is as follows: Edge nodes preprocess data and upload key information to the cloud; The AI ​​model is trained in the cloud and then distributed to edge nodes to update the local model.

[0024] Preferably, the real-time parameters of the equipment are mapped to the BIM model in real time through a 5G private network to form a dynamic digital twin; By connecting the BIM API with a dynamic digital twin database, real-time parameter-driven status updates of the BIM model can be achieved.

[0025] Preferably, the real-time parameters of the turbine include bearing vibration, guide vane opening, head height, and flow rate; A turbine fault prediction model is built. Real-time parameters of the turbine are input, an LSTM neural network model is used to predict the remaining life of the bearing, an anomaly detection algorithm is used to identify abnormal vibration modes, and potential faults and recommended maintenance solutions are output 7 days in advance.

[0026] Preferably, underwater inspection equipment equipped with a 5G camera is used to capture videos of the inner wall of the pipeline. The YOLOv8+U-Net joint model is used to detect cracks and cavitation areas and quantify the damage area. The detection results are automatically labeled to the BIM model to generate repair work orders.

[0027] Preferably, a reinforcement learning-based scheduling model is built to optimize the pumping / power generation switching strategy based on the grid load demand and reservoir water level. The input parameters of the reinforcement learning-based scheduling model include electricity price, water level and unit efficiency curve, dynamically generating the optimal scheduling command, which is then sent to the unit control module via the 5G network.

[0028] Preferably, it also includes a project progress prediction module, which includes project progress prediction and work order management functions; The project progress forecasting function specifically includes: dynamically adjusting the project schedule based on the overall construction progress, predicting construction steps in advance, preparing materials, personnel and equipment, and generating construction priority work orders; The work order management function specifically includes: automatically dispatching work orders to the responsible person's mobile terminal and using AR to remotely guide installation and maintenance procedures.

[0029] Preferably, it also includes an emergency command module, which simulates extreme working conditions based on a BIM model, and uses AI to predict the impact range of a dam failure and generate evacuation routes; When the backbone network is interrupted, 5G D2D devices are used to directly connect and build a temporary communication network to ensure the transmission of command instructions.

[0030] Preferably, network slicing includes control command slices, video surveillance slices, and BIM data slices; The control command slice carries unit start / stop commands and valve control commands; Video surveillance slices carry video from drone inspections or images or video data collected by underwater inspection equipment; BIM data slices carry dynamic data of the BIM model and AR / VR interactive data.

[0031] Preferably, the hardware configuration and tasks of the edge computing nodes are obtained, the edge computing nodes are divided into three levels, different levels are configured with different amounts of BIM models, and the tasks are set to the highest priority to ensure real-time response. Different scales of BIM models refer to lightweight BIM models and lightweight local AI models that include only modules related to the work task and parameters related to the work task and match the hardware configuration of the current node. Lightweight BIM models refer to BIM models that minimize redundant data and have lower precision around the work task. Lightweight local AI models refer to models that are used only for data preprocessing related to work tasks or for executing tasks issued by cloud computing centers without iterative training.

[0032] Preferably, the distributed training and inference process of the LSTM neural network model includes: The real-time acquired turbine bearing vibration signals are preprocessed to extract key features; a lightweight LSTM model is used for real-time inference to predict the bearing's remaining life RUL(t). The inference formula is as follows: ; in, For LSTM hidden states, For input features, For LSTM inference functions; Historical data from several edge nodes are aggregated, and a federated learning framework is used for global model training. Edge nodes upload model gradients, and the cloud aggregates and updates the global model parameters. ; in, This represents the global model parameters after aggregation in the cloud. These are the initial parameters for the local model. Where is the learning rate, and N is the number of edge nodes. The model gradient calculated for the i-th edge node based on local data; The cloud will update the global model parameters The model is deployed to edge nodes to replace the local model and optimize inference accuracy.

[0033] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A 5G+BIM intelligent interactive management system for pumped storage power stations, characterized by: include: The perception layer is used to collect equipment implementation parameters, including real-time parameters of water turbines, generators, water pipelines and reservoir water levels, and to realize dynamic data updates of the BIM model through 5G cameras, sensors and underwater inspection equipment. At the network layer, dynamic data of the BIM model is transmitted through the 5G private network, and communication resources are allocated according to business needs using a network slicing strategy. The data layer is used to integrate dynamic data from the BIM model, including BIM geometric data, real-time equipment parameters, and historical operation and maintenance data, to build a dynamic digital twin database. At the application layer, predictive maintenance, intelligent scheduling, and risk warning are performed based on equipment parameters and BIM models, and decision instructions are generated through deep learning algorithms. The edge-cloud collaborative computing module is used to dynamically allocate computing resources according to the real-time requirements of the task, through distributed training and inference of LSTM neural network model and reinforcement learning-based scheduling model. Edge computing nodes are deployed in 5G base stations, field servers, or terminals to perform tasks with high real-time requirements. Edge computing nodes are divided into several levels based on their assessed computing power, with different levels corresponding to different computing tasks; The cloud computing center provides computing tasks of a corresponding scale based on the edge computing nodes, and feeds the results back to the edge nodes through a 5G private network; The specific mechanism of edge-cloud collaboration is as follows: Edge nodes preprocess data and upload key information to the cloud; The AI ​​model is trained in the cloud and then distributed to edge nodes to update the local model.

2. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: The equipment's real-time parameters are mapped to the BIM model in real time through a 5G private network, forming a dynamic digital twin; By connecting the BIM API with a dynamic digital twin database, real-time parameter-driven status updates of the BIM model can be achieved.

3. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: Real-time parameters of a water turbine include bearing vibration, guide vane opening, head height, and flow rate; A turbine fault prediction model is built. Real-time parameters of the turbine are input, an LSTM neural network model is used to predict the remaining life of the bearing, an anomaly detection algorithm is used to identify abnormal vibration modes, and potential faults and recommended maintenance solutions are output 7 days in advance.

4. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: Underwater inspection equipment equipped with a 5G camera is used to capture videos of the inner wall of the pipeline. Using a YOLOv8+U-Net joint model, cracks and cavitation areas are detected and the damage area is quantified. The detection results are automatically labeled to the BIM model and a repair work order is generated.

5. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: A reinforcement learning-based scheduling model was built to optimize the pumping / generation switching strategy based on grid load demand and reservoir water level. The input parameters of the reinforcement learning-based scheduling model include electricity price, water level and unit efficiency curve. The optimal scheduling command is dynamically generated and sent to the unit control module through the 5G network.

6. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: It also includes a project progress forecasting module, which includes project progress forecasting and work order management functions; The project progress forecasting function specifically includes: dynamically adjusting the project schedule based on the overall construction progress, predicting construction steps in advance, preparing materials, personnel and equipment, and generating construction priority work orders; The work order management function specifically includes: automatically dispatching work orders to the responsible person's mobile terminal and using AR to remotely guide installation and maintenance procedures.

7. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: It also includes an emergency command module, which simulates extreme working conditions based on BIM models and uses AI to predict the impact range of dam failure and generate evacuation routes; When the backbone network is interrupted, 5G D2D devices are used to directly connect and build a temporary communication network to ensure the transmission of command instructions.

8. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: Network slicing includes control command slices, video surveillance slices, and BIM data slices; The control command slice carries unit start / stop commands and valve control commands; Video surveillance slices carry video from drone inspections or images or video data collected by underwater inspection equipment; BIM data slices carry dynamic data of the BIM model and AR / VR interactive data.

9. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: Obtain the hardware configuration and tasks of the edge computing nodes, divide the edge computing nodes into three levels, configure different levels of BIM models, and set the tasks to the highest priority to ensure real-time response; Different scales of BIM models refer to lightweight BIM models and lightweight local AI models that include only modules related to the work task and parameters related to the work task and match the hardware configuration of the current node. Lightweight BIM models refer to BIM models that minimize redundant data and have lower precision around the work task. Lightweight local AI models refer to models that are used only for data preprocessing related to work tasks or for executing tasks issued by cloud computing centers without iterative training.

10. The 5G+BIM intelligent interactive management system for pumped storage power stations according to claim 1, characterized in that: The distributed training and inference process of an LSTM neural network model includes: The real-time acquired turbine bearing vibration signals are preprocessed to extract key features; a lightweight LSTM model is used for real-time inference to predict the bearing's remaining life RUL(t). The inference formula is as follows: ； in, For LSTM hidden states, For input features, For LSTM inference functions; Historical data from several edge nodes are aggregated, and a federated learning framework is used for global model training. Edge nodes upload model gradients, and the cloud aggregates and updates the global model parameters. ； in, This represents the global model parameters after aggregation in the cloud. These are the initial parameters for the local model. Where is the learning rate, and N is the number of edge nodes. The model gradient calculated for the i-th edge node based on local data; The cloud will update the global model parameters The model is deployed to edge nodes to replace the local model and optimize inference accuracy.

Images