A power station intelligent data decision analysis visualization method and system

Abstract

The application discloses a power station intelligent data decision analysis visualization method and system, and belongs to the technical field of power station intelligent data decision analysis, and comprises the following steps: constructing a multi-source data acquisition system, deploying the system on power station equipment nodes, environment monitoring points and power grid interaction ends, collecting equipment operation parameters, environment sensing data, electric energy metering data and fault history data in real time, and forming an original data set. Through three-dimensional rendering and intelligent analysis deep fusion, the pain points of traditional static display and lack of trend presentation are solved, the panoramic view is associated with the spatial position and operation state of the equipment, the dynamic flowchart intuitively displays load changes and scheduling paths, and through high-light marking of the fault equipment, hierarchical alarm pop-up windows and sound and light reminders, the decision response time is shortened, and core information can be quickly mastered without professional data analysis capability. The technical problem that the prior art lacks deep fusion with an intelligent analysis model and thus cannot dynamically perform visualization is solved.

Description

Technical Field

[0001] This application belongs to the field of power plant intelligent data decision analysis technology, specifically involving a power plant intelligent data decision analysis visualization method and system. Background Technology

[0002] During the operation of a power plant, a massive amount of heterogeneous data from multiple sources is generated, including equipment operating parameters, environmental sensing data, power metering data, and historical fault data. The effective integration, intelligent analysis, and intuitive presentation of this data are the core prerequisites for achieving precise operation and maintenance and optimized load scheduling of the power plant.

[0003] Existing methods and systems for power plant data visualization have significant shortcomings: First, visualization often remains at the level of static data display, lacking deep integration with intelligent analysis models and failing to dynamically present equipment health status, fault risks, and load change trends. Second, data preprocessing is not precise enough, with poor compatibility of heterogeneous data protocols and inconsistent timestamps leading to distorted visualized data. Third, interactive functions are limited, failing to support multi-dimensional operations such as hierarchical penetration and parameter filtering, making it difficult to meet the needs of refined queries. Fourth, system response and refresh strategies are fixed, failing to dynamically adjust according to data importance, resulting in poor balance between real-time performance and operational efficiency, and lacking a closed-loop mechanism for continuous optimization, making it difficult to adapt to the complex and ever-changing operating scenarios of power plants. Summary of the Invention

[0004] This application addresses the technical problem that existing technologies lack deep integration with intelligent analysis models, thus making dynamic visualization impossible. It provides a method and system for visualizing intelligent data decision analysis in power plants.

[0005] To achieve the above objectives, this application adopts the following technical solution: The first aspect of this application is a method for visualizing intelligent data decision analysis in power plants, comprising: Collect equipment operating parameters, environmental sensing data, power metering data, and fault history data of the power station to form a raw dataset; Outliers in the original dataset are removed, and timestamp consistency calibration and standardization are performed to obtain a clean dataset; The clean dataset is input into a pre-trained intelligent analysis model to obtain load prediction results, equipment failure risk probability, and health status score; the intelligent analysis model consists of a short-term load prediction model and an equipment health assessment model. Based on a geographic information system, the device's operating parameters and health status scores are correlated with spatial location to generate a panoramic visualization base view. When the probability of equipment failure risk is greater than or equal to a preset threshold, the corresponding equipment will be highlighted and alarmed in the panoramic visualization base view. A dynamic flowchart will be generated based on the load prediction results. The dynamic flowchart is used to display the load change trend and load scheduling path. Dynamic visualization results will be generated based on the panoramic visualization base view and the dynamic flowchart.

[0006] In some implementations, the step of generating the dynamic visualization result is followed by: Obtain power plant operation procedures and load dispatch requirements; Based on the power plant operation procedures and load scheduling requirements, and based on the dynamic visualization results, decision information is generated, which is then converted into executable control signals and transmitted to the power plant control system. The data refresh frequency of the dynamic visualization results is dynamically adjusted, and user operation data and system operation feedback data are collected. The interface layout of the dynamic visualization results is adjusted based on the user operation data, and the parameters of the intelligent analysis model are updated based on the system operation feedback data.

[0007] In some implementations, the short-term load forecasting model includes: a double-stacked LSTM layer, an attention mechanism layer, and a fully connected output layer; The double-stacked LSTM layer is used to extract the hidden state vector of each time step in the clean dataset, and output the temporal dependency feature sequence based on the hidden state vector; The attention mechanism layer is used to adaptively weight the hidden state vectors at each time step based on the temporal dependent feature sequence to generate a context vector. The fully connected output layer is used to output load prediction results based on the context vector.

[0008] In some implementations, the device health assessment model includes: a feature extraction module and a deep neural network, wherein the deep neural network includes an input layer, a hidden layer and an output layer; The feature extraction module is used to extract trend features and correlation features from the device operating parameters; The deep neural network is used to input the trend features and correlation features into the input layer and transmit them to the output layer through the hidden layer. The output layer outputs the device failure risk probability through the Sigmoid activation function and linearly maps the device failure risk probability to a health status score.

[0009] In some implementations, training the pre-trained intelligent analysis model includes: Obtain the historical clean dataset and divide it into training and validation sets according to time series. The training set is used to iteratively train the short-term load prediction model and the equipment health assessment model, respectively. The short-term load prediction model uses the root mean square error as the loss function and the Adam optimizer for parameter updates. The equipment health assessment model uses the historical fault data as labels for supervised training. The trained short-term load prediction model and equipment health assessment model are validated and evaluated using the validation set to obtain the pre-trained intelligent analysis model.

[0010] In some implementations, removing outliers from the original dataset and performing timestamp consistency calibration and standardization to obtain a clean dataset includes: Outliers in the original dataset were removed based on the 3σ principle, timestamp consistency was calibrated using the NTP protocol, and finally, the dataset was standardized using the Z-Score standardization formula to obtain a clean dataset.

[0011] A second aspect of this application discloses a power plant intelligent data decision analysis and visualization system, comprising: The data acquisition module is used to collect equipment operating parameters, environmental sensing data, power metering data and fault history data of the power station to form a raw dataset; The data preprocessing module is used to remove outliers from the original dataset and perform timestamp consistency calibration and standardization to obtain a clean dataset. The intelligent analysis module is used to input the cleaning dataset into a pre-trained intelligent analysis model to obtain load prediction results, equipment failure risk probability, and health status score; the intelligent analysis model consists of a short-term load prediction model and an equipment health assessment model. The visualization view generation module is used to associate the device's operating parameters and health status scores with spatial location based on the geographic information system to generate a panoramic visualization base view. The dynamic visualization module is used to highlight and alarm the corresponding device in the panoramic visualization base view when the probability of device failure risk is greater than or equal to a preset threshold, and generate a dynamic flowchart based on the load prediction results. The dynamic flowchart is used to display the load change trend and load scheduling path. The dynamic visualization result is generated according to the panoramic visualization base view and the dynamic flowchart.

[0012] A third aspect of this application is a computer device comprising: a processor and a computer-readable storage medium; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the aforementioned intelligent data decision analysis and visualization method for power plants.

[0013] A fourth aspect of this application is a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed as described in the power plant intelligent data decision analysis and visualization method.

[0014] A fifth aspect of this application is a computer program product comprising a computer program that, when executed by a processor, implements the aforementioned intelligent data decision analysis and visualization method for power plants.

[0015] Compared with the prior art, this application has the following beneficial effects: This application collects equipment operating parameters, environmental sensing data, electricity metering data, and historical fault data to form a raw dataset. The raw dataset is then subjected to outlier removal, timestamp consistency calibration, and standardization to obtain a clean dataset. This ensures the integrity, temporal consistency, and format uniformity of the input data, providing a reliable data foundation for subsequent analysis. The clean dataset is input into a pre-trained intelligent analysis model, which consists of a short-term load forecasting model and an equipment health assessment model. This model can simultaneously output load forecasting results, equipment fault risk probability, and health status scores. Based on a geographic information system, the equipment operating parameters and health status scores are correlated with spatial location to generate a panoramic visualization base view, allowing the equipment status to be intuitively presented in a spatial dimension. When the equipment fault risk probability is greater than or equal to a preset threshold, the corresponding equipment is highlighted in the panoramic visualization base view and an alarm is triggered. Simultaneously, a dynamic flowchart displaying load change trends and load scheduling paths is generated based on the load forecasting results, ultimately producing a dynamic visualization result. The above process tightly couples the output of the intelligent analysis model with the visualization presentation. The probability of fault risk directly drives the highlighting and alarm actions, and the load prediction results directly drive the generation of dynamic flowcharts. It establishes a real-time linkage mechanism from data analysis to the visualization interface, which solves the problem of the disconnect between visualization and intelligent analysis in the existing technology and the inability to dynamically reflect the equipment status and load trends. This enables maintenance personnel to quickly identify abnormal equipment and load change trends without the need for professional data analysis background, thus shortening the decision response time.

[0016] Furthermore, the system acquires power plant operation procedures and load scheduling requirements, using these as constraints for decision generation. Decision information is generated based on dynamic visualization results and transformed into executable control signals transmitted to the power plant control system. This allows the visualization analysis conclusions to directly impact the actual operation and control of the power plant, forming a closed-loop decision-making process from discovery to execution. The system dynamically adjusts the data refresh frequency of the visualization results and simultaneously collects user operation data and system operation feedback data. The interactive layout of the visualization interface is adjusted based on user operation data to better suit user habits. The parameters of the intelligent analysis model are updated based on system operation feedback data, enabling the model to continuously fit actual operating conditions during operation. This process not only completes single analysis and display but also achieves continuous iterative optimization of interface interaction and model accuracy through dual feedback from both the user and system sides, forming a complete closed loop of data acquisition, analysis, visualization, decision-making, and optimization. This ensures decision accuracy and operational convenience during long-term operation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart of a power plant intelligent data decision analysis visualization method provided in this application embodiment; Figure 2 A structural diagram of a power plant intelligent data decision analysis and visualization system provided in this application embodiment; Figure 3 The data preprocessing and intelligent analysis sub-flowchart provided for the embodiments of this application; Figure 4 A flowchart illustrating the visualization rendering and interactive decision-making process provided in this application embodiment; Figure 5 The following is a flowchart of the closed-loop optimization and system response adjustment sub-flow provided in the embodiments of this application; Figure 6 A structural diagram of a computer device provided for an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] like Figure 1 As shown, this embodiment provides a visualization method for intelligent data decision analysis in power plants, including: S1 collects equipment operating parameters, environmental sensing data, power metering data, and fault history data of the power station to form the raw dataset; Specifically, a multi-source data acquisition system is constructed and deployed at power plant equipment nodes, environmental monitoring points and power grid interaction terminals. Through acquisition terminals adapted to Modbus, IEC61850 and MQTT protocols, equipment operating parameters, environmental sensing data, power metering data and fault history data are collected in real time to form raw datasets. To address the heterogeneous network environment within the power plant, the data acquisition terminal incorporates a multi-protocol adaptation module. For core production equipment (such as generator sets and transformers), a Modbus-to-TCP / IP gateway is used to read data from the device's internal registers via polling. For distributed environmental monitoring points (such as temperature, humidity, and wind speed sensors), low-power wireless sensor nodes supporting the MQTT protocol (such as LoRaWAN or NB-IoT) are used to publish data to the central MQTTBroker. For metering devices interacting with the upstream power grid, a smart electronic device (IED) interface supporting the IEC61850 standard is used for data acquisition. The acquisition terminal hardware is based on an embedded ARM architecture, and the software runs a lightweight real-time operating system to ensure the stability and real-time performance of data acquisition.

[0021] The equipment operating parameters comprehensively cover key indicators reflecting equipment status, such as voltage, current, power factor, active / reactive power, frequency, winding temperature, and bearing vibration values. Environmental sensing data includes external factors affecting equipment operating efficiency and safety, such as ambient temperature and humidity, wind speed, precipitation, atmospheric pressure, and air quality index (AQI). Fault history data is obtained by connecting to the power plant operation and maintenance management system database, and structured extraction of equipment fault records from the past 3-5 years is performed, including fault type, occurrence time, alarm information, handling process, repair time, and post-recovery operating parameters, providing key samples for subsequent model training.

[0022] The data acquisition terminal has a built-in data buffer queue. When the network connection with the central server is interrupted, the data can be temporarily stored locally and resumed after the network is restored, ensuring data integrity. All acquired data is accompanied by a high-precision timestamp at the source end and is synchronized with the central server at the sub-millisecond level via NTP (Network Time Protocol), providing a unified time reference for subsequent multi-source data fusion analysis.

[0023] Equipment operating parameters include voltage, current, power factor, winding temperature, and equipment vibration values. Environmental sensing data includes ambient temperature and humidity, wind speed, precipitation, and air quality index. Fault history data covers equipment fault types, occurrence times, handling processes, and recovery parameters for the past 3-5 years. By adapting to mainstream industrial communication protocols in power plant scenarios such as Modbus, IEC61850, and MQTT, the system addresses data heterogeneity issues between different devices, monitoring points, and the power grid system, achieving data interconnection and interoperability. Data acquisition terminals are strategically deployed at equipment nodes, environmental monitoring points, and power grid interaction points to accurately capture four key data categories: equipment operating parameters, environmental sensing data, electricity metering data, and fault history data (full fault information for the past 3-5 years). This ensures the comprehensiveness and real-time nature of the raw dataset, providing complete and reliable basic data support for subsequent analysis and preventing distortion of analysis results due to missing or outdated data.

[0024] S2, Remove outliers from the original dataset, and perform timestamp consistency calibration and standardization to obtain a clean dataset; Specifically, the original dataset is preprocessed, outliers are removed based on the 3σ principle, timestamp consistency calibration is performed through the NTP protocol, and heterogeneous data is transformed into a unified format using a data standardization algorithm to generate a clean dataset. The data standardization algorithm employs the Z-Score standardization formula to address issues affecting the accuracy of analysis, such as outlier interference, inconsistent timestamps, and heterogeneous formats in the original data. It utilizes a standardized processing flow to purify and unify the data, accurately identifying and removing outliers that deviate from the normal distribution based on the 3σ principle, ensuring data authenticity. The NTP protocol is used to calibrate the timestamp consistency of multi-source data, resolving the issue of time asynchrony between data from different acquisition terminals and ensuring data temporal correlation. The Z-Score standardization formula transforms heterogeneous data of different magnitudes and units into standardized data with a mean of 0 and a standard deviation of 1, eliminating the influence of dimensions and generating a clean dataset with a uniform format and reliable quality, providing high-quality input for intelligent model analysis.

[0025] S3, input the cleaning dataset into the pre-trained intelligent analysis model to obtain load prediction results, equipment failure risk probability and health status score; the intelligent analysis model consists of a short-term load prediction model and an equipment health assessment model; Specifically, an intelligent analysis model is built, which integrates the LSTM+Attention short-term load prediction model and the equipment health assessment model. The clean dataset is input for training and iteration, and the load prediction results, equipment failure risk probability and health status score are output. The short-term load forecasting model structure includes: an input layer that receives processed multidimensional time-series data (historical load, weather data, date type, etc.); two stacked LSTM (Long Short-Term Memory) layers, each containing 128 neurons, to capture long-term dependencies in the load data; an attention mechanism layer that calculates weights for different time steps, enabling the model to focus on historical data points that have the greatest impact on the forecast results; and a fully connected output layer that outputs the load forecast value for a specified future time step. The model is trained using the Adam optimizer, and the loss function is the root mean square error (RMSE).

[0026] Therefore, the short-term load prediction model includes: a double-stacked LSTM layer, an attention mechanism layer, and a fully connected output layer; the double-stacked LSTM layer is used to extract the hidden state vectors of each time step in the clean dataset and output a time-dependent feature sequence based on the hidden state vectors; the attention mechanism layer is used to perform adaptive weighted calculation on the hidden state vectors of each time step based on the time-dependent feature sequence to generate a context vector; the fully connected output layer is used to output the load prediction result based on the context vector.

[0027] The equipment health assessment model is used to quantitatively evaluate the health status of critical equipment and predict failure risks. The model construction process includes: first, extracting trend features (such as moving averages and growth rates) and correlation features (such as the correlation between current and temperature) from equipment operating parameters; then, constructing a deep neural network (DNN) with four hidden layers. The input layer receives the extracted features, and the output layer outputs a failure risk probability between 0 and 1 through a sigmoid activation function, linearly converting it into a health status score of 0-100. The model is trained under supervised supervision using historical failure data, and a score below 60 is defined as a risk warning threshold.

[0028] The equipment health assessment model includes: a feature extraction module and a deep neural network, wherein the deep neural network includes an input layer, a hidden layer and an output layer; The feature extraction module is used to extract trend features and correlation features from the device operating parameters; The deep neural network is used to input the trend features and correlation features into the input layer and transmit them to the output layer through the hidden layer. The output layer outputs the device failure risk probability through the Sigmoid activation function and linearly maps the device failure risk probability to a health status score.

[0029] The equipment health assessment model extracts trend features and fault correlation features of equipment operating parameters, constructs a multi-layer neural network model, and outputs a health status score of 0-100 after training with historical fault data. A risk warning is triggered when the score is below 60.

[0030] The training of the pre-trained intelligent analysis model includes: Obtain the historical clean dataset and divide it into training and validation sets according to time series. The training set is used to iteratively train the short-term load prediction model and the equipment health assessment model, respectively. The short-term load prediction model uses the root mean square error as the loss function and the Adam optimizer for parameter updates. The equipment health assessment model uses the historical fault data as labels for supervised training. The trained short-term load prediction model and equipment health assessment model are validated and evaluated using the validation set to obtain the pre-trained intelligent analysis model.

[0031] like Figure 3 As shown, the LSTM+Attention short-term load forecasting model leverages the strong fitting ability of LSTM to time-series data to capture the patterns of load data changes over time. Simultaneously, the Attention mechanism strengthens the weight of data at key time points, improving the accuracy of load forecasting. The equipment health assessment model extracts trend features and fault correlation features of equipment operating parameters, constructs a multi-layer neural network model, and forms a quantitative assessment capability for equipment health status, outputting a health score of 0-100 and a fault risk probability. This enables accurate judgment of equipment health status and fault early warning. The two models work together to output load forecasting results, fault risk probabilities, and health status scores, providing core analytical basis for subsequent visualization and decision-making.

[0032] S4. Based on the geographic information system, the device's operating parameters and health status score are associated with spatial location to generate a panoramic visualization base view; Specifically, a three-dimensional visualization rendering engine is built based on a GIS geographic information system, which integrates the topology of power plant equipment with geospatial information, and associates equipment operating parameters, health status scores and spatial locations to generate a panoramic visualization base view; The 3D visualization rendering engine employs a hybrid rendering scheme combining JavaFX and WebGL. Leveraging the spatial positioning capabilities of a GIS (Geographic Information System), and combining JavaFX and WebGL, it constructs a high-performance 3D visualization rendering engine. This engine deeply integrates the power plant equipment topology with geospatial information, achieving precise correlation mapping between equipment operating parameters, health status scores, and spatial location.

[0033] S5, when the probability of equipment failure risk is greater than or equal to a preset threshold, the corresponding equipment is highlighted and alarmed in the panoramic visualization base view. A dynamic flowchart is generated based on the load prediction result. The dynamic flowchart is used to display the load change trend and load scheduling path. A dynamic visualization result is generated according to the panoramic visualization base view and the dynamic flowchart.

[0034] Specifically, the visualization view is dynamically updated based on the intelligent analysis results. Devices with a fault risk probability exceeding the preset threshold are highlighted and alarm pop-ups are displayed. The load change trend and scheduling optimization path are presented intuitively through dynamic flowcharts. The alarm pop-up includes the probability of failure risk, information on associated devices, estimated failure time, and preliminary handling suggestions. At the same time, it also alerts maintenance personnel through an audible and visual alarm device. The alarm level is divided into three levels: general, emergency, and special, based on the probability of risk.

[0035] The system monitors the comparison between the probability of equipment failure risk and preset thresholds in real time. Equipment exceeding the threshold is automatically highlighted and a pop-up alarm window is generated, containing the probability of failure risk, related equipment information, estimated failure time, and preliminary handling suggestions. The system also activates audible and visual alarm devices, classifying alarms into three levels: general, emergency, and special, based on the probability of failure, to ensure that maintenance personnel can respond quickly to potential faults. For load change trends and scheduling optimization paths, the system presents them intuitively through dynamic flowcharts, visualizing the temporal patterns and logical relationships of data changes. This solves the problem that traditional static displays cannot reflect real-time status and trends, allowing users to keep track of the power plant's operational dynamics in real time.

[0036] The following steps are included after S5: S6: Provides multi-dimensional interactive functions, supports view zooming, hierarchical penetration, parameter filtering and historical data backtracking query. Users can adjust the parameter thresholds of the analysis model through the interactive interface, trigger secondary analysis and update the visualization results in real time. S7: Generates a decision suggestion library based on the results of visualization analysis, and outputs equipment maintenance priorities, power allocation schemes and fault emergency handling procedures in combination with power plant operation procedures and load dispatching requirements. Simultaneously, it is converted into executable control signals and fed back to the power plant control system. S8: Dynamically adjusts the data refresh frequency through a system response speed optimization algorithm. Core operating parameters are refreshed at the millisecond level, while non-core parameters are adaptively reduced in refresh frequency, balancing real-time visualization and system operating efficiency. Specifically, the system response speed optimization algorithm described in S8 dynamically adjusts the refresh frequency based on the device's operating status. The refresh frequency of the core device is 50ms / time when it is running normally, and it is increased to 20ms / time when it is in a fault state. The refresh frequency of the auxiliary device is maintained at 500ms / time.

[0037] By adjusting the data refresh frequency through system response speed optimization algorithms, the real-time visualization and system operating efficiency are balanced, and non-core data is prevented from consuming too many system resources. Through differentiated refresh strategies, the system load is reduced and the overall operating stability is improved while meeting the real-time requirements of core data.

[0038] S9: Collect user operation data and system operation feedback data, continuously optimize intelligent analysis model parameters and visualization interface interaction logic, and form a closed-loop mechanism of data collection, analysis, visualization, decision-making and optimization.

[0039] like Figure 4 As shown, steps S4 to S6 in this embodiment involve constructing a 3D visualization rendering engine and implementing dynamic interaction. The specific implementation method is as follows: A hybrid rendering scheme using JavaFX and WebGL is adopted. The backend uses JavaFX to handle complex business logic, data analysis, and scene management, while the frontend loads a 3D scene based on WebGL (implemented through the Three.js library) through the built-in WebView component. JavaFX communicates with the WebGL scene through a bidirectional JavaScript interface, pushing backend analysis results (such as device status and alarm information) to the frontend in real time for visualization updates, while also receiving user interaction events from the frontend.

[0040] The integration of GIS and BIM, based on the macro-geographic coordinates of the power plant provided by the GIS geographic information system, combined with the equipment-level 3D models and topology provided by BIM (Building Information Modeling), achieves precise spatial mapping from the macro-geographic environment to the micro-details of the equipment. The position, size, and orientation of each equipment model in the 3D scene are consistent with the actual location, and it serves as a data carrier, associated with its unique equipment ID.

[0041] Dynamic interactive functionality is implemented, with layer-by-layer penetration triggered by mouse double-clicks or scroll wheel zoom events. The system dynamically loads models at different levels of detail (LOD) based on the distance between the current viewpoint and the equipment model, enabling seamless drilling from a panoramic view of the power plant to equipment groups, and then to the internal structure of individual equipment (such as transformer windings). Parameter filtering and historical data retrieval are triggered through controls on the interface (such as checkboxes, sliders, and date pickers). After receiving the filtering conditions, the JavaFX backend executes a query in the database and pushes the result set to the frontend, dynamically updating the highlighted equipment or historical data curves in the scene.

[0042] The layer penetration function described in step S6 allows you to penetrate layer by layer from the panoramic view of the power plant to the detailed view of a single device, displaying the device's real-time operating parameters, historical data curves, and health status change trends. The parameter filtering function supports precise filtering by device type, operating period, and data threshold.

[0043] The view zoom function allows users to adjust the visualization range according to their observation needs, taking into account both the overall overview and the viewing of local details; the layer penetration function uses the view layer drill-down mechanism to penetrate from the overall view of the power plant to the details of a single device, displaying the real-time parameters, historical data curves and health status change trends of the device, meeting the needs of refined analysis from the whole to the part.

[0044] like Figure 5 As shown, steps S7 to S9 in this embodiment of the invention generate decision suggestions and form a closed-loop optimization, and the specific implementation method is as follows: Decision suggestion library generation: The decision suggestion library is built on an expert rule engine. The engine has a built-in rule set that combines power plant operation procedures and historical fault cases. For example, when the temperature of transformer A exceeds 90°C for 10 consecutive minutes and its health score is below 65, a level 1 maintenance work order is generated. When the intelligent analysis results meet the trigger conditions of specific rules, the engine automatically generates equipment maintenance priorities, power allocation suggestions, or fault emergency handling procedures, and converts them into control instructions that comply with the IEC 61850 standard.

[0045] System response speed optimization: This algorithm is implemented through a dynamic priority queue. Operating parameters of core equipment, such as the main transformer and generator, are assigned high priority, and their data refresh requests have priority processing rights in the queue, achieving refresh frequencies of 50ms or even 20ms. Auxiliary equipment or non-critical parameters are assigned low priority, and their refresh requests wait in the queue or adopt an adaptive frequency reduction strategy, thereby effectively reducing the overall system load while ensuring the real-time performance of core data.

[0046] Closed-loop optimization mechanism: The system backend continuously records user interface operation logs (such as click frequency and view dwell time) and system operation feedback data (such as model prediction accuracy and alarm response time). Every quarter, by analyzing this data offline, the weight parameters of the intelligent analysis model are fine-tuned using the gradient descent algorithm to optimize the accuracy of prediction and evaluation. At the same time, based on heatmap analysis of user operation habits, the menu layout and interaction logic of the visualization interface are adjusted, placing frequently used functions in more convenient locations, thereby continuously improving decision analysis efficiency and user experience.

[0047] The closed-loop mechanism described in step S9 optimizes the weight parameters of the intelligent analysis model through the gradient descent algorithm, adjusts the menu layout and interaction logic of the visualization interface based on user operating habits, and completes a full-process optimization iteration every quarter to improve the accuracy of decision analysis and the usability of the interface.

[0048] By collecting user operation data and system operation feedback data, we can obtain system optimization directions, use the gradient descent algorithm to optimize the weight parameters of the intelligent analysis model, improve the accuracy of load forecasting, health assessment and fault early warning, ensure long-term stable improvement of system performance, and continuously optimize decision analysis results and user experience.

[0049] In one embodiment of this application, such as Figure 2 As shown, a power plant intelligent data decision analysis and visualization system is provided, including: The data acquisition module is used to collect equipment operating parameters, environmental sensing data, power metering data and fault history data of the power station to form a raw dataset; The data preprocessing module is used to remove outliers from the original dataset and perform timestamp consistency calibration and standardization to obtain a clean dataset. The intelligent analysis module is used to input the cleaning dataset into a pre-trained intelligent analysis model to obtain load prediction results, equipment failure risk probability, and health status score; the intelligent analysis model consists of a short-term load prediction model and an equipment health assessment model. The visualization view generation module is used to associate the device's operating parameters and health status scores with spatial location based on the geographic information system to generate a panoramic visualization base view. The dynamic visualization module is used to highlight and alarm the corresponding device in the panoramic visualization base view when the probability of device failure risk is greater than or equal to a preset threshold. It generates a dynamic flowchart based on the load prediction results. The dynamic flowchart is used to display the load change trend and load scheduling path. The dynamic visualization result is generated based on the panoramic visualization base view and the dynamic flowchart.

[0050] Specific limitations regarding a power plant intelligent data decision analysis and visualization system can be found in the above description of a power plant intelligent data decision analysis and visualization method; the corresponding technical effects are equivalent and will not be repeated here. Each module in the aforementioned power plant intelligent data decision analysis and visualization system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0051] Figure 6 An internal structural diagram of a computer device is shown in one embodiment. This computer device may specifically be a terminal or a server. Figure 6As shown, the computer device includes a processor, memory, network interface, display, camera, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a power plant intelligent data decision analysis and visualization method. The display screen can be an LCD screen or an e-ink display screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0052] As will be understood by those skilled in the art, computer equipment Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device to which the present invention is applied. A specific computing device may include more or fewer components than those shown in the figure, or combine certain components, or have the same component arrangement.

[0053] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0054] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0055] In summary, the intelligent data decision analysis visualization method, system, computer equipment, and storage medium provided in this application address the pain point of traditional static displays lacking trend presentation through deep integration of 3D rendering and intelligent analysis. A panoramic view links the spatial location and operating status of equipment, while dynamic flowcharts intuitively display load changes and scheduling paths. Simultaneously, highlighted faulty equipment, tiered alarm pop-ups, and audio-visual alerts shorten decision response time, allowing for rapid understanding of core information without requiring specialized data analysis capabilities. Outlier removal using the 3σ principle, timestamp calibration via NTP protocol, and Z-Score standardization ensure data consistency and accuracy. The LSTM+Attention model exhibits low load prediction error, improves fault risk identification accuracy, and reduces unplanned downtime losses. Multi-dimensional interaction supports view zooming, hierarchical penetration, parameter filtering, and historical backtracking to meet refined query needs. A dynamic refresh strategy adjusts the frequency as needed, with core parameters updated in milliseconds and non-core parameters adaptively reduced in frequency, balancing real-time performance and operational efficiency. This adapts to complex power plant scenarios and reduces overall operation and maintenance costs.

[0056] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0057] The above-described embodiments are merely preferred embodiments of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. A visualization method for intelligent data decision analysis in power plants, characterized in that, include: Collect equipment operating parameters, environmental sensing data, power metering data, and fault history data of the power station to form a raw dataset; Outliers in the original dataset are removed, and timestamp consistency calibration and standardization are performed to obtain a clean dataset; The clean dataset is input into a pre-trained intelligent analysis model to obtain load prediction results, equipment failure risk probability, and health status score. The intelligent analysis model consists of a short-term load forecasting model and an equipment health assessment model. Based on a geographic information system, the device's operating parameters and health status scores are correlated with spatial location to generate a panoramic visualization base view. When the probability of equipment failure risk is greater than or equal to a preset threshold, the corresponding equipment will be highlighted and alarmed in the panoramic visualization base view. A dynamic flowchart will be generated based on the load prediction results. The dynamic flowchart is used to display the load change trend and load scheduling path. Dynamic visualization results will be generated based on the panoramic visualization base view and the dynamic flowchart.

2. The intelligent data decision analysis and visualization method for power plants according to claim 1, characterized in that, Following the step of generating dynamic visualization results, the following steps are included: Obtain power plant operation procedures and load dispatch requirements; Based on the power plant operation procedures and load scheduling requirements, and based on the dynamic visualization results, decision information is generated, which is then converted into executable control signals and transmitted to the power plant control system. The data refresh frequency of the dynamic visualization results is dynamically adjusted, and user operation data and system operation feedback data are collected. The interface layout of the dynamic visualization results is adjusted based on the user operation data, and the parameters of the intelligent analysis model are updated based on the system operation feedback data.

3. The intelligent data decision analysis and visualization method for power plants according to claim 1, characterized in that, The short-term load forecasting model includes: a double-stacked LSTM layer, an attention mechanism layer, and a fully connected output layer; The double-stacked LSTM layer is used to extract the hidden state vector of each time step in the clean dataset, and output the temporal dependency feature sequence based on the hidden state vector; The attention mechanism layer is used to adaptively weight the hidden state vectors at each time step based on the temporal dependent feature sequence to generate a context vector. The fully connected output layer is used to output load prediction results based on the context vector.

4. The intelligent data decision analysis and visualization method for power plants according to claim 1, characterized in that, The equipment health assessment model includes: a feature extraction module and a deep neural network, wherein the deep neural network includes an input layer, a hidden layer and an output layer; The feature extraction module is used to extract trend features and correlation features from the device operating parameters; The deep neural network is used to input the trend features and correlation features into the input layer and transmit them to the output layer through the hidden layer. The output layer outputs the device failure risk probability through the Sigmoid activation function and linearly maps the device failure risk probability to a health status score.

5. The intelligent data decision analysis and visualization method for power plants according to claim 1, characterized in that, The training of the pre-trained intelligent analysis model includes: Obtain the historical clean dataset and divide it into training and validation sets according to time series. The training set is used to iteratively train the short-term load prediction model and the equipment health assessment model, respectively. The short-term load prediction model uses the root mean square error as the loss function and the Adam optimizer for parameter updates. The equipment health assessment model uses the historical fault data as labels for supervised training. The trained short-term load prediction model and equipment health assessment model are validated and evaluated using the validation set to obtain the pre-trained intelligent analysis model.

6. The intelligent data decision analysis and visualization method for power plants according to claim 1, characterized in that, The process of removing outliers from the original dataset and performing timestamp consistency calibration and standardization to obtain a clean dataset includes: Outliers in the original dataset were removed based on the 3σ principle, timestamp consistency was calibrated using the NTP protocol, and finally, the dataset was standardized using the Z-Score standardization formula to obtain a clean dataset.

7. A power plant intelligent data decision analysis and visualization system, characterized in that, include: The data acquisition module is used to collect equipment operating parameters, environmental sensing data, power metering data and fault history data of the power station to form a raw dataset; The data preprocessing module is used to remove outliers from the original dataset and perform timestamp consistency calibration and standardization to obtain a clean dataset. The intelligent analysis module is used to input the cleaning dataset into a pre-trained intelligent analysis model to obtain load prediction results, equipment failure risk probability, and health status score. The intelligent analysis model consists of a short-term load forecasting model and an equipment health assessment model. The visualization view generation module is used to associate the device's operating parameters and health status scores with spatial location based on the geographic information system to generate a panoramic visualization base view. The dynamic visualization module is used to highlight and alarm the corresponding device in the panoramic visualization base view when the probability of device failure risk is greater than or equal to a preset threshold. It generates a dynamic flowchart based on the load prediction results. The dynamic flowchart is used to display the load change trend and load scheduling path. The dynamic visualization result is generated based on the panoramic visualization base view and the dynamic flowchart.

8. A computer device, characterized in that, include: Processor and computer-readable storage media; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements a power plant intelligent data decision analysis and visualization method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements a power plant intelligent data decision analysis and visualization method as described in any one of claims 1 to 6.

Images