Method and system for real-time positioning and trajectory analysis of wind power plant workers

Abstract

The invention belongs to the technical field of wind power plant management, and relates to a method and system for real-time positioning and trajectory analysis of wind power plant workers. The method comprises the following steps of: preprocessing multi-source position data by utilizing an edge computing node; fusing the preprocessed multi-source position data to generate a high-precision positioning result; performing trajectory analysis by using a deep learning algorithm to obtain a trajectory analysis result; performing intelligent task allocation by combining wind power plant equipment state data and environment data; on the basis of the physiological data and the environmental data of the staff, health condition monitoring and safety risk assessment of the staff are carried out; generating an equipment maintenance strategy by using a predictive maintenance algorithm in combination with the trajectory analysis result and the wind power plant equipment state data; and a worker real-time trajectory analysis result, an intelligent task allocation result, a monitoring and evaluation result and a standby maintenance strategy are visually presented. The positioning precision is improved, track analysis is accurate and reasonable, real-time safety monitoring of workers can be achieved, and the intelligent degree is high.

Description

Technical Field

[0001] The invention belongs to the technical field of wind farm management, and relates to a method and system for real-time positioning and trajectory analysis of wind farm workers. Background Art

[0002] Real-time positioning and trajectory analysis of wind farm workers is a key technology for wind farm safety management and efficient operation. This technology involves real-time tracking of the positions of wind farm workers, analyzing their movement trajectories, and performing safety monitoring, task allocation, and equipment maintenance optimization based on this information. Through real-time positioning and trajectory analysis, the operational efficiency of wind farms can be effectively improved, safety risks can be reduced, and resource allocation can be optimized.

[0003] At present, the positioning of wind farm workers mainly relies on the Global Positioning System or other single positioning technologies. The Global Positioning System is referred to as GPS. However, these methods often have difficulty in ensuring the accuracy and reliability of positioning in complex wind farm environments. For example, inside tall wind turbine towers or in areas with complex terrain, GPS signals may be severely interfered with or completely lost. In addition, existing trajectory analysis methods usually use simple statistical analysis or rule-based methods, which makes it difficult to deeply understand the behavior patterns and work efficiency of workers, and the trajectory analysis is not deep enough. In terms of safety management, most wind farms still rely on traditional manual inspections and fixed safety plans, lack dynamic monitoring of the real-time status of workers and intelligent risk warning mechanisms, and safety monitoring is not intelligent enough.

[0004] In summary, the existing trajectory analysis methods have problems such as insufficient positioning accuracy, insufficient trajectory analysis depth, and insufficient intelligence in safety monitoring. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for real-time positioning and trajectory analysis of wind farm workers, so as to solve the technical problems of insufficient positioning accuracy, insufficient trajectory analysis depth, and insufficient intelligence of safety monitoring in existing wind farm trajectory analysis methods. The present invention comprehensively utilizes multi-source position data to improve positioning accuracy, accurate and reasonable trajectory analysis, and can realize real-time safety monitoring of workers with a high degree of intelligence.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for real-time positioning and trajectory analysis of wind farm workers, comprising the following steps: Obtain multi-source location data of wind farm workers, physiological data of workers, status data of wind farm equipment and environmental data; Preprocessing the multi-source location data using edge computing nodes; Fusion of pre-processed multi-source location data generates high-precision positioning results; Based on the high-precision positioning result, a trajectory analysis is performed using a deep learning algorithm to obtain a trajectory analysis result; Intelligent task allocation is performed based on trajectory analysis results combined with wind farm equipment status data and environmental data; Based on the physiological and environmental data of the workers, the health status of the workers is monitored and the safety risk is assessed to obtain the monitoring and assessment results; Use predictive maintenance algorithms to combine trajectory analysis results and wind farm equipment status data to generate equipment maintenance strategies; Visualize the real-time trajectory analysis results, intelligent task allocation results, monitoring and evaluation results, and maintenance strategies of staff.

[0007] In a second aspect, the present invention provides a system for real-time positioning and trajectory analysis of wind farm workers, comprising: Data acquisition module: used to obtain multi-source location data of wind farm workers, physiological data of workers, wind farm equipment status data and environmental data; Data preprocessing module: used to preprocess the multi-source location data using edge computing nodes; Positioning result acquisition module: used to fuse pre-processed multi-source location data to generate high-precision positioning results; Trajectory analysis module: used to perform trajectory analysis based on the high-precision positioning result using a deep learning algorithm to obtain a trajectory analysis result; Task allocation module: used to perform intelligent task allocation based on trajectory analysis results combined with wind farm equipment status data and environmental data; Safety risk assessment module: used to monitor the health status of workers and conduct safety risk assessment based on their physiological data and environmental data, and obtain monitoring and assessment results; Equipment maintenance strategy generation module: used to generate equipment maintenance strategies by combining the predictive maintenance algorithm with the trajectory analysis results and the wind farm equipment status data; Visualization presentation module: used to visualize the real-time trajectory analysis results of staff, intelligent task allocation results, monitoring and evaluation results, and maintenance strategies.

[0008] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention provides a comprehensive information basis for subsequent analysis and processing by acquiring multi-source location data of wind farm workers, physiological data of workers, status data of wind farm equipment and environmental data. The multi-source location data is preprocessed by edge computing nodes. By preprocessing the multi-source location data, the accuracy and reliability of the positioning results can be improved. The high-precision positioning results are generated by fusing the preprocessed multi-source location data, which is conducive to improving the positioning accuracy. Based on the high-precision positioning results, trajectory analysis is performed using a deep learning algorithm to obtain trajectory analysis results, which is conducive to optimizing the workflow and improving production efficiency. According to the trajectory analysis results, combined with wind farm equipment status data and environmental data, intelligent task allocation is performed, which helps to improve work efficiency and reduce resource waste. Based on the physiological data and environmental data of the workers, the health status of the workers is monitored and the safety risk assessment is performed, and the monitoring and assessment results are obtained, which can timely discover potential health problems and safety risks. This helps to protect the life safety of the workers and reduce the occurrence of work-related accidents. The predictive maintenance algorithm is combined with the trajectory analysis results and the wind farm equipment status data to generate equipment maintenance strategies; the real-time trajectory analysis results of the staff, the intelligent task allocation results, the monitoring and evaluation results and the maintenance strategies are visualized, which helps managers to intuitively understand the overall operation of the wind farm and make more informed decisions. The present invention comprehensively utilizes multi-source location data to improve positioning accuracy, accurate and reasonable trajectory analysis, and can achieve real-time safety monitoring of staff, with a high degree of intelligence.

[0009] 2. The system of the present invention includes: a data acquisition module, a data preprocessing module, a positioning result acquisition module, a trajectory analysis module, a task allocation module, a safety risk assessment module, an equipment maintenance strategy generation module and a visualization module; the data acquisition module is used to acquire multi-source location data of wind farm workers, physiological data of workers, wind farm equipment status data and environmental data; the data preprocessing module is used to preprocess the multi-source location data using edge computing nodes; the positioning result acquisition module is used to fuse the preprocessed multi-source location data to generate high-precision positioning results; the trajectory analysis module is used to perform trajectory analysis based on the high-precision positioning results using a deep learning algorithm to obtain trajectory analysis results; the task allocation module is used to perform intelligent task allocation based on the trajectory analysis results in combination with the wind farm equipment status data and environmental data; the safety risk assessment module is used to monitor the health status of workers and conduct safety risk assessment based on the physiological data and environmental data of workers to obtain monitoring and assessment results; the equipment maintenance strategy generation module is used to generate equipment maintenance strategies using a predictive maintenance algorithm in combination with trajectory analysis results and wind farm equipment status data; the visualization module is used to visualize the real-time trajectory analysis results of workers, intelligent task allocation results, monitoring and assessment results and standby maintenance strategies. The various modules cooperate with each other and improve the positioning accuracy through the comprehensive use of multi-source location data. The trajectory analysis is accurate and reasonable, and real-time safety monitoring of staff can be achieved with a high degree of intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A flow chart of a method for real-time positioning and trajectory analysis of wind farm workers according to an embodiment of the present invention; Figure 2 This is an overall architecture diagram of a real-time positioning and trajectory analysis system for wind farm workers according to an embodiment of the present invention; Figure 3 This is a multi-source data fusion positioning flow chart of an embodiment of the present invention; Figure 4 A structural diagram of a safety management and health monitoring system according to an embodiment of the present invention; Figure 5 Flow chart of the method of the present invention; Figure 6 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0011] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0012] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0013] The present invention is further described in detail below in conjunction with the accompanying drawings: See also Figure 5 The present invention discloses a method for real-time positioning and trajectory analysis of wind farm workers, comprising the following steps: S1. Obtain multi-source location data of wind farm workers, physiological data of workers, wind farm equipment status data and environmental data; S2. Preprocess the multi-source location data using edge computing nodes, as follows: Use wavelet transform technology to remove noise from multi-source position data; Use the isolation forest algorithm to detect outliers in the multi-source location data after noise removal and delete the abnormal data; The Douglas-Peucker algorithm is used to compress the multi-source location data after deleting abnormal data. The compression rate is greater than 90% and the accuracy loss is less than 1%.

[0014] S3. Fusion of pre-processed multi-source location data generates high-precision positioning results, as follows: The multi-source position data includes: global navigation satellite system data, inertial measurement unit data, Bluetooth beacon data and ultra-wideband positioning data; The extended Kalman filter algorithm is used to perform tight coupling fusion of global navigation satellite system data and inertial measurement unit data to obtain tightly coupled fusion data; Beacon positioning is performed using particle filtering algorithm and Bluetooth beacon data; Use multi-lateration positioning algorithm and ultra-wideband positioning data for ultra-wideband positioning; High-precision positioning results are generated by fusing tightly coupled fusion data, beacon positioning results and ultra-wideband positioning results through the federated Kalman filter.

[0015] S4. Based on the high-precision positioning results, a deep learning algorithm is used to perform trajectory analysis to obtain trajectory analysis results, which are as follows: The basic high-precision positioning results are segmented using the spatiotemporal density clustering algorithm; According to the results of trajectory segmentation, the behavior recognition is performed in combination with the hierarchical long short-term memory network to obtain the behavior recognition result; The trajectory prediction is performed using a deep learning model based on the self-attention mechanism and a graph attention network model combined with the behavior recognition results to obtain the trajectory analysis results.

[0016] S5. Based on the trajectory analysis results, combined with the wind farm equipment status data and environmental data, intelligent task allocation is performed as follows: Use a hierarchical task network to perform task modeling and obtain a modeling task; Use a multi-dimensional scoring model to dynamically evaluate staff capabilities and obtain capability scores; Based on the deep reinforcement learning method of the proximal policy optimization algorithm, the task allocation is optimized by combining the modeling tasks, trajectory analysis results, capability scores, current tasks and equipment status to achieve intelligent task allocation; Use graph cut algorithm to perform dynamic team division; An adaptive time division multiple access protocol based on task urgency is used for communication scheduling.

[0017] S6. Based on the physiological data and environmental data of the staff, the health status of the staff and the safety risk assessment are carried out to obtain the monitoring and assessment results, as follows: Use the improved YOLOv5 model to obtain personal protective equipment inspection results; Build microclimate models based on environmental data combined with computational fluid dynamics and machine learning; Assess fatigue based on heart rate variability data from workers’ physiological data; The heat stress risk index is obtained based on the microclimate model combined with the temperature change trend in the physiological data of the workers; Monitor the health status of workers and conduct safety risk assessment based on personal protective equipment inspection results, fatigue level and heat stress risk index, and obtain monitoring and assessment results; If the monitoring and evaluation results are unqualified, a multi-level alarm is implemented based on the rule engine, and the improved A-star algorithm is used to plan the optimal evacuation route.

[0018] Preferably, the improved YOLOv5 model is used, specifically as follows: The improved YOLOv5 model is used to introduce a spatial attention module into the backbone network; The improved YOLOv5 model adopts an adaptive feature fusion strategy to improve the feature pyramid network.

[0019] The improved YOLOv5 model is described to achieve model lightweighting through channel pruning and knowledge distillation technology.

[0020] Preferably, the improved A-star algorithm is used to plan the optimal evacuation path, specifically as follows: Adjust the path cost function weights based on real-time environmental data and monitoring and evaluation results; The optimal evacuation path is planned based on the path cost function, multi-objective optimization strategy, hierarchical search strategy and fuzzy logic processing method.

[0021] S7. Generate equipment maintenance strategy by combining trajectory analysis results and wind farm equipment status data using a predictive maintenance algorithm; S8. Visualize the real-time trajectory analysis results, intelligent task allocation results, monitoring and evaluation results, and maintenance strategies of workers.

[0022] The present invention significantly improves the accuracy of positioning and trajectory analysis by integrating multi-source positioning data and advanced deep learning algorithms. This method can maintain stable high-precision positioning in complex wind farm environments, including inside wind turbine towers and areas with complex terrain, effectively overcoming the limitations of traditional single positioning technology. The intelligent trajectory analysis and task allocation mechanism greatly improves work efficiency and optimizes human resource allocation. At the same time, the real-time health monitoring and safety risk assessment system significantly enhances the safety management level of wind farms and effectively reduces the safety risks of workers. The predictive maintenance function optimizes maintenance strategies, extends equipment life, and reduces unplanned downtime by combining worker trajectory data and equipment status information.

[0023] In addition, the application of quantum encryption communication technology ensures the security of data transmission, while the visual presentation of the digital twin platform provides an intuitive and real-time management interface.

[0024] In general, the present invention comprehensively improves the operational efficiency, safety and sustainability of wind farms, and provides strong technical support for the intelligent management of wind farms.

[0025] See also Figure 5In another feasible embodiment of the present invention, the following is adaptively modified according to the situation. It includes the following steps: Acquiring multi-source location data of wind farm workers, physiological data of workers, wind farm equipment status data and environmental data provides a comprehensive information basis for subsequent analysis and processing.

[0026] The multi-source location data is preprocessed using edge computing nodes. By preprocessing the multi-source location data, the accuracy and reliability of the positioning results can be improved.

[0027] Fusion of preprocessed multi-source location data generates high-precision positioning results, which is beneficial to improving positioning accuracy.

[0028] Based on the high-precision positioning results, trajectory analysis is performed using a deep learning algorithm to obtain trajectory analysis results, which is conducive to optimizing the workflow and improving production efficiency.

[0029] Intelligent task allocation based on trajectory analysis results combined with wind farm equipment status data and environmental data helps improve work efficiency and reduce resource waste.

[0030] Based on the physiological and environmental data of workers, the health status of workers is monitored and safety risk is assessed. The monitoring and assessment results can timely identify potential health problems and safety risks. This helps to protect the lives of workers and reduce the occurrence of work-related accidents.

[0031] Use predictive maintenance algorithms to combine trajectory analysis results and wind farm equipment status data to generate equipment maintenance strategies; Visual presentation of real-time trajectory analysis results, intelligent task allocation results, monitoring and evaluation results, and maintenance strategies of staff members helps managers to intuitively understand the overall operation of the wind farm and make more informed decisions.

[0032] Embodiment 1: This embodiment provides a method for real-time positioning and trajectory analysis of wind farm workers, wherein: Figure 2 This is the overall architecture diagram of the real-time positioning and trajectory analysis system for wind farm workers. The main steps of this method are as follows: Figure 1 As shown, including: Step S1: Data collection. The smart work clothes are used to collect the location data of wind farm workers, including global navigation satellite system (GNSS) data, inertial measurement unit (IMU) data, Bluetooth beacon data, and ultra-wideband (UWB) positioning data.

[0033] Data collection is mainly completed through smart work clothes that integrate multiple sensors. The smart work clothes include a high-precision multi-mode global navigation satellite system receiver, a nine-axis inertial measurement unit, a low-power Bluetooth 5.2 module, an ultra-wideband positioning module, a 5G communication module, a biosensor kit, and an environmental sensor. The high-precision multi-mode global navigation satellite system receiver is referred to as GNSS, the nine-axis inertial measurement unit is referred to as IMU, and the ultra-wideband positioning module is referred to as UWB positioning module. This diversified sensor configuration ensures the comprehensiveness and accuracy of data collection. The GNSS receiver supports GPS, Beidou, Galileo and GLONASS systems, with a sampling rate of 10Hz, providing a basis for outdoor positioning. The IMU contains a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer, with a sampling rate of up to 100Hz, providing key data for inertial navigation and attitude estimation. The UWB positioning module is specifically used for high-precision indoor positioning, and its 200Hz sampling rate ensures millimeter-level positioning accuracy.

[0034] Step S2: Data preprocessing: Use edge computing nodes to preprocess the collected raw data, including data cleaning, outlier detection, and data compression.

[0035] The edge computing node is responsible for the preliminary processing of the collected raw data, which mainly includes three steps: noise removal, outlier detection and data compression. Noise removal uses wavelet transform technology, uses db4 wavelet basis function to perform three-layer decomposition on GNSS data, and removes high-frequency noise through soft threshold method.

[0036] Outlier detection uses the Isolation Forest algorithm, which has the advantage of processing high-dimensional data and low computational complexity. The calculation formula for the anomaly score is as follows:

[0037] in, For sample The anomaly score, For sample The average path length, For a given The average path length for samples is is the sample size.

[0038] The Douglas-Peucker algorithm is used for data compression. By setting a suitable distance threshold, a compression rate of over 90% is achieved while keeping the accuracy loss within 1%.

[0039] Step S3: Data transmission: The pre-processed data is securely transmitted to the central control system through quantum encryption communication technology.

[0040] Step S4: Multi-source data fusion positioning: Multi-source location data is fused in the central control system to generate high-precision positioning results.

[0041] like Figure 3 As shown in the figure, the multi-source data fusion positioning process includes the following steps: First, GNSS data and IMU data are initially fused through GNSS / IMU tight coupling. At the same time, Bluetooth beacon data is processed by particle filter positioning, and UWB data is processed by multilateral positioning. Finally, these processing results are input into the federal Kalman filter to generate the final high-precision positioning result, which can achieve centimeter-level accuracy. This multi-source data fusion method effectively improves the accuracy and reliability of positioning, especially in complex wind farm environments.

[0042] The central control system receives the pre-processed data and achieves high-precision positioning through sophisticated algorithms. The GNSS / IMU tight coupling adopts the extended Kalman filter (EKF) algorithm, whose state vector includes position, velocity, attitude quaternion, accelerometer bias and gyroscope bias, and the observation vector includes GNSS pseudorange, carrier phase and Doppler shift. Bluetooth beacon positioning combines weighted centroid positioning and particle filter algorithms, while UWB positioning uses two-way ranging (TWR) technology and multilateral positioning algorithms. Finally, various positioning results are fused through the federal Kalman filter to achieve centimeter-level positioning accuracy.

[0043] Step S5: Trajectory analysis: Based on the high-precision positioning results, deep learning algorithms are used to perform trajectory analysis, including trajectory segmentation, behavior recognition, and trajectory prediction.

[0044] Based on the high-precision positioning results, the present invention uses deep learning technology to perform trajectory analysis. Trajectory analysis includes three main steps: trajectory segmentation, behavior recognition and trajectory prediction. Trajectory segmentation uses the ST-DBSCAN algorithm for spatiotemporal clustering to divide continuous trajectory data into discrete behavior segments. The spatiotemporal density clustering algorithm is referred to as ST-DBSCAN, and its full English name is Spatio-Temporal Density-Based Spatial Clustering of Applications with Noise. Behavior recognition uses a hierarchical long short-term memory network, which is referred to as LSTM. The network structure includes a low-level LSTM for extracting temporal features, a high-level LSTM for capturing long-term dependencies, and finally outputs the behavior category probability through a fully connected layer and a Softmax layer. Trajectory prediction uses a combination model of Transformer and graph attention network. Transformer is used to capture temporal features, and GAT is used to model spatial relationships. Transformer is a deep learning model based on the self-attention mechanism, and the attention network is referred to as GAT.

[0045] The loss function of the trajectory prediction model is as follows:

[0046] in, is the mean square error loss, which is used to evaluate the position prediction accuracy; is the cross entropy loss, which is used to evaluate the accuracy of behavior prediction; is a regularization term used to prevent overfitting; , and is the weight coefficient, is the loss function of the trajectory prediction model.

[0047] Step S6: Intelligent task allocation: Based on the trajectory analysis results, combined with the wind farm equipment status and environmental data, intelligent task allocation and collaborative optimization are performed.

[0048] The present invention realizes intelligent task allocation and collaborative optimization. Based on the trajectory analysis results, the system combines the wind farm equipment status and environmental data to realize intelligent task allocation and team collaboration.

[0049] Task modeling adopts hierarchical task network, abbreviated as HTN, which takes into account multiple attributes such as task priority, required skills, estimated time and location constraints.

[0050] Workforce competency assessment uses a multi-dimensional scoring model to dynamically evaluate skill proficiency, physical condition, and historical completion rates.

[0051] Task allocation optimization adopts a deep reinforcement learning method based on the proximal policy optimization algorithm. Its state space includes the worker's location, ability score, current task and equipment status. The action space is the task allocation decision, and the reward function comprehensively considers the task completion rate, efficiency improvement and load balancing.

[0052] Team collaboration optimization uses a graph cut algorithm to dynamically divide teams, and uses an adaptive time division multiple access protocol based on task urgency for communication scheduling. The proximal policy optimization algorithm is referred to as PPO, and the adaptive time division multiple access protocol is referred to as TDMA.

[0053] Step S7: Safety monitoring and risk assessment: Based on the physiological data and environmental data of the staff, health status monitoring and safety risk assessment are performed.

[0054] like Figure 4 As shown in the figure, the safety management and health monitoring system includes multiple interrelated components. The system realizes real-time health status monitoring and safety risk assessment based on the physiological data and environmental data of the staff. The main components of the system include: physiological data collection and environmental data collection provide basic data; fatigue assessment, heat stress risk warning and PPE compliance inspection are carried out for specific analysis; microclimate model provides support for heat stress risk warning; comprehensive risk assessment integrates various analysis results; finally, according to the evaluation results, a multi-level alarm mechanism is triggered or the optimal evacuation route is planned.

[0055] Fatigue assessment analyzes heart rate variability data and calculates the ratio of low-frequency components (LF) to high-frequency components (HF) LF / HF as a fatigue indicator. Heat stress risk warning combines body temperature change trends, ambient temperature and humidity data, and uses a modified physiological heat stress index for assessment. The system also establishes a wind farm microclimate model based on computational fluid dynamics and machine learning to predict local weather changes. Among them, heart rate variability data is referred to as HRV, physiological heat stress index is referred to as mPHS, and computational fluid dynamics is referred to as CFD.

[0056] YOLOv5 is a computer vision model known for real-time object detection. It is an upgraded version of the YOLO series and was developed by Ultralytics. The personal protective equipment inspection uses an improved YOLOv5 model to achieve real-time PPE compliance inspection. The improvement is mainly reflected in the following aspects: First, for the special environment of wind farms, specific data enhancement techniques are used, such as simulating different weather conditions, lighting changes and occlusion conditions, to improve the recognition ability of the model in complex environments. Secondly, the spatial attention module is introduced in the backbone network of YOLOv5 to enhance the model's ability to capture detailed features and improve the detection accuracy of small targets (such as helmets and goggles). Thirdly, the feature pyramid network (FPN) is improved, and an adaptive feature fusion strategy is adopted to better handle PPE targets of different scales. Finally, through channel pruning and knowledge distillation technology, the model calculation complexity is reduced while maintaining high accuracy, and real-time inference on edge devices is achieved. Among them, personal protective equipment is referred to as PPE and feature pyramid network is referred to as FPN.

[0057] In an emergency, the system implements a multi-level alarm mechanism based on a rule engine and uses an improved A-star algorithm to plan the optimal evacuation path. In this invention, the standard A-star algorithm is improved to better adapt to the complex environment and emergency needs of wind farms. The A-star algorithm is referred to as A-star. The improved A-star algorithm mainly includes the following innovations: 1. Dynamic weight adjustment: The traditional A-star algorithm uses a fixed cost function, while our improved version dynamically adjusts the weight of the path cost function based on real-time environmental data (such as wind speed, temperature) and worker status (such as fatigue). Specifically, the cost function f(n) is modified as follows: f(n) = g(n) + w(e,s) * h(n) Where g(n) is the actual cost from the starting point to the current node n, h(n) is the estimated cost from the current node to the target, and w(e,s) is a weight factor dynamically calculated based on the environment (e) and the person's state (s).

[0058] 2. Multi-objective optimization: The improved algorithm not only considers the path length, but also optimizes safety and time factors at the same time. We adopted a variant of the multi-objective A-star algorithm, using a Pareto optimal solution set to provide multiple options for decision making. This allows the system to select the most appropriate evacuation path based on the specific situation in an emergency.

[0059] 3. Hierarchical search strategy: To handle the path planning problem of large-scale wind farms, we adopted the hierarchical A-star algorithm. The wind farm area is divided into multiple sub-areas, firstly roughly planned at the high level, and then refined at the low level. This method significantly improves the efficiency of large-scale path planning.

[0060] 4. Fuzzy logic integration: In order to handle the uncertainty in the wind farm environment, fuzzy logic is introduced in the A-star algorithm. For example, the boundary of obstacles may not be clear, and we use fuzzy sets to represent this uncertainty and consider this fuzzy information in path evaluation.

[0061] This improved A-star algorithm can quickly calculate the optimal evacuation path in the complex environment of a wind farm, effectively improving the response speed and safety in emergency situations. By combining the A-star algorithm with other safety management system components, the present invention significantly enhances the safety level of wind farm workers.

[0062] Step S8: Predictive maintenance: Use the predictive maintenance algorithm to combine the worker trajectory data and equipment status data to generate equipment maintenance strategies.

[0063] The predictive maintenance system optimizes equipment maintenance strategies by innovatively combining worker trajectory data and equipment status data. Worker trajectory data provides spatiotemporal information on equipment use and inspection, while equipment status data directly reflects the health of the equipment. Equipment health assessment uses a one-dimensional convolutional neural network (1D-CNN) and an autoencoder to fuse vibration data, temperature, power curves, and audio features to output equipment health index and probability distribution of fault types. Remaining useful life (RUL) prediction uses a long short-term memory network (LSTM) combined with a Monte Carlo dropout method, performs feature engineering through wavelet packet decomposition, and extracts statistical features as model input. The maintenance scheduling optimization uses the multi-objective genetic algorithm NSGA-III, which comprehensively considers reliability maximization, maintenance cost minimization, and optimal allocation of human resources, while considering constraints such as weather windows, spare parts availability, and technician skill matching.

[0064] Step S9: Data visualization: Visualize the real-time location and trajectory information of the staff and related analysis results on the digital twin platform.

[0065] The present invention successfully solves several key problems in the real-time positioning and trajectory analysis of wind farm workers by innovatively integrating multiple advanced technologies, including multi-source data fusion, deep learning, reinforcement learning, quantum encryption communication, etc. This method not only achieves high-precision positioning and intelligent trajectory analysis, but also provides comprehensive safety management, health monitoring and predictive maintenance functions. Through this comprehensive technical solution, the present invention significantly improves the operational efficiency and safety of wind farms, and provides strong technical support for the intelligent management of wind farms. The implementation of the present invention can not only improve the economic benefits of wind farms, but also greatly improve the safety of workers. It has broad application prospects and important practical value.

[0066] Based on the above method, the present invention discloses a system for real-time positioning and trajectory analysis of wind farm workers, see Figure 6 ,include: Data acquisition module: used to obtain multi-source location data of wind farm workers, physiological data of workers, wind farm equipment status data and environmental data; Data preprocessing module: used to preprocess the multi-source location data using edge computing nodes; Positioning result acquisition module: used to fuse pre-processed multi-source location data to generate high-precision positioning results; Trajectory analysis module: used to perform trajectory analysis based on the high-precision positioning result using a deep learning algorithm to obtain a trajectory analysis result; Task allocation module: used to perform intelligent task allocation based on trajectory analysis results combined with wind farm equipment status data and environmental data; Safety risk assessment module: used to monitor the health status of workers and conduct safety risk assessment based on their physiological data and environmental data, and obtain monitoring and assessment results; Equipment maintenance strategy generation module: used to generate equipment maintenance strategies by combining the predictive maintenance algorithm with the trajectory analysis results and the wind farm equipment status data; Visualization presentation module: used to visualize the real-time trajectory analysis results of staff, intelligent task allocation results, monitoring and evaluation results, and maintenance strategies.

[0067] The various modules of the system of the present invention cooperate with each other, and through the comprehensive use of multi-source position data, the positioning accuracy is improved, the trajectory analysis is accurate and reasonable, and the real-time safety monitoring of the staff can be realized, with a high degree of intelligence.

[0068] The present invention can realize high-precision real-time positioning, intelligent trajectory analysis and dynamic safety monitoring in complex wind farm environments. It should be able to effectively solve the problems of insufficient positioning accuracy, insufficient trajectory analysis depth, and insufficient intelligence of safety monitoring in the existing technology, thereby comprehensively improving the operational efficiency and safety management level of wind farms.

[0069] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for real-time positioning and trajectory analysis of wind farm workers, characterized in that: The following steps are involved: Obtain multi-source location data of wind farm workers, physiological data of workers, status data of wind farm equipment and environmental data; Preprocessing the multi-source location data using edge computing nodes; Fusion of pre-processed multi-source location data generates high-precision positioning results; Based on the high-precision positioning result, a trajectory analysis is performed using a deep learning algorithm to obtain a trajectory analysis result; Intelligent task allocation is performed based on trajectory analysis results combined with wind farm equipment status data and environmental data; Based on the physiological and environmental data of the staff, the health status of the staff and the safety risk assessment are carried out to obtain the monitoring and assessment results; Use predictive maintenance algorithms to combine trajectory analysis results and wind farm equipment status data to generate equipment maintenance strategies; Visualize the real-time trajectory analysis results, intelligent task allocation results, monitoring and evaluation results, and maintenance strategies of staff.

2. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: The edge computing node is used to pre-process the multi-source location data, specifically as follows: Use wavelet transform technology to remove noise from multi-source position data; Use the isolation forest algorithm to detect outliers in the multi-source location data after noise removal and delete the abnormal data; The Douglas-Peucker algorithm is used to compress the multi-source location data after deleting abnormal data, and the compression rate is greater than 90%.

3. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: The fusion preprocessed multi-source position data generates a high-precision positioning result, as follows: The multi-source position data includes: global navigation satellite system data, inertial measurement unit data, Bluetooth beacon data and ultra-wideband positioning data; The extended Kalman filter algorithm is used to perform tight coupling fusion of global navigation satellite system data and inertial measurement unit data to obtain tightly coupled fusion data; Beacon positioning is performed using particle filtering algorithm and Bluetooth beacon data; Use multi-lateration positioning algorithm and ultra-wideband positioning data for ultra-wideband positioning; High-precision positioning results are generated by fusing tightly coupled fusion data, beacon positioning results and ultra-wideband positioning results through the federated Kalman filter.

4. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: Based on the high-precision positioning result, the trajectory analysis is performed using a deep learning algorithm to obtain a trajectory analysis result, which is as follows: The basic high-precision positioning results are segmented using the spatiotemporal density clustering algorithm; According to the results of trajectory segmentation, the behavior recognition is performed in combination with the hierarchical long short-term memory network to obtain the behavior recognition result; The trajectory prediction is performed using a deep learning model based on the self-attention mechanism and a graph attention network model combined with the behavior recognition results to obtain the trajectory analysis results.

5. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: According to the trajectory analysis results, combined with the wind farm equipment status data and environmental data, intelligent task allocation is performed as follows: Use a hierarchical task network to perform task modeling and obtain a modeling task; Use a multi-dimensional scoring model to dynamically evaluate staff capabilities and obtain capability scores; The deep reinforcement learning method based on the proximal policy optimization algorithm combines the modeling tasks, trajectory analysis results, capability scores, current tasks and equipment status to optimize task allocation and realize intelligent task allocation.

6. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: Based on the physiological data and environmental data of the staff, the health status of the staff and the safety risk assessment are carried out to obtain the monitoring and assessment results, which are as follows: Use the improved YOLOv5 model to obtain personal protective equipment inspection results; Build microclimate models based on environmental data combined with computational fluid dynamics and machine learning; Assess fatigue based on heart rate variability data from workers’ physiological data; Obtain heat stress risk index based on microclimate model combined with temperature change trend in physiological data of workers; Monitor the health status of workers and conduct safety risk assessment based on personal protective equipment inspection results, fatigue level and heat stress risk index, and obtain monitoring and assessment results; If the monitoring and evaluation results are unqualified, a multi-level alarm is implemented based on the rule engine, and the improved A-star algorithm is used to plan the optimal evacuation route.

7. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: The improved YOLOv5 model is used as follows: The improved YOLOv5 model is used to introduce a spatial attention module into the backbone network; The improved YOLOv5 model adopts an adaptive feature fusion strategy to improve the feature pyramid network.

8. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: The improved YOLOv5 model is used as follows: The improved YOLOv5 model is described to achieve model lightweighting through channel pruning and knowledge distillation technology.

9. The method for real-time positioning and trajectory analysis of wind farm workers according to claim 1, characterized in that: The improved A-star algorithm is used to plan the optimal evacuation path, as follows: Adjust the path cost function weights based on real-time environmental data and monitoring and evaluation results; The optimal evacuation path is planned based on the path cost function, multi-objective optimization strategy, hierarchical search strategy and fuzzy logic processing method.

10. A system for real-time positioning and trajectory analysis of wind farm workers, characterized in that: include: Data acquisition module: used to obtain multi-source location data of wind farm workers, physiological data of workers, wind farm equipment status data and environmental data; Data preprocessing module: used to preprocess the multi-source location data using edge computing nodes; Positioning result acquisition module: used to fuse pre-processed multi-source location data to generate high-precision positioning results; Trajectory analysis module: used to perform trajectory analysis based on the high-precision positioning result using a deep learning algorithm to obtain a trajectory analysis result; Task allocation module: used to perform intelligent task allocation based on trajectory analysis results combined with wind farm equipment status data and environmental data; Safety risk assessment module: used to monitor the health status of workers and conduct safety risk assessment based on their physiological data and environmental data, and obtain monitoring and assessment results; Equipment maintenance strategy generation module: used to generate equipment maintenance strategies by combining the predictive maintenance algorithm with the trajectory analysis results and the wind farm equipment status data; Visualization presentation module: used to visualize the real-time trajectory analysis results of staff, intelligent task allocation results, monitoring and evaluation results, and maintenance strategies.

Images