Coal mine personnel positioning and safety early warning method

Abstract

The invention relates to the technical field of mine safety, in particular to a coal mine personnel positioning and safety early warning method, which comprises the following steps: acquiring position information of coal mine personnel; processing the position information to obtain high-precision positioning data; analyzing the high-precision positioning data through a clustering algorithm, and marking potential safety hazard data points; performing anomaly detection on the potential safety hazard data points by adopting an isolated forest algorithm to obtain abnormal behavior characteristics and potential safety hazards; and performing risk assessment based on the abnormal behavior characteristics and the potential safety hazards to obtain a risk assessment value, comparing the risk assessment value with a preset risk threshold, and triggering safety early warning according to a comparison result. According to the invention, accurate positioning, real-time monitoring and abnormal behavior early warning of the coal mine personnel can be realized, and the coal mine safety management level and the accident prevention capability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of mine safety, and particularly to a method for positioning coal mine personnel and safety warning. Background Art

[0002] As an important energy industry field, the coal mine production process involves a large number of underground operations, and the working environment is relatively closed and complex. The safety of coal mine personnel has always been one of the most important topics in the coal mine production process. There are many uncertain factors in the underground environment of coal mines, such as gas leakage, roof fall, cave-in, etc., which may even seriously threaten the lives of miners in severe cases. Therefore, how to effectively locate coal mine workers and timely detect potential dangers, and then take necessary safety warning measures, is the key to ensuring coal mine production safety.

[0003] Due to the complex and changeable coal mine environment, there are a large number of obstacles and interference sources, and traditional single positioning technologies are difficult to meet the requirements of precise positioning; at the same time, different positioning technologies perform differently in different application scenarios. Therefore, comprehensively applying multiple positioning technologies and giving full play to their respective advantages is the key technical issue for achieving all-round high-precision positioning of coal mine personnel. In addition, due to the harsh coal mine environment, communication signals are easily interfered with and attenuated, making it difficult to ensure the real-time, reliability, and security of positioning data; at the same time, a large amount of positioning data needs to be efficiently stored, analyzed, and mined, and interconnected and shared with other processing parts. Therefore, data transmission and processing in the coal mine environment also face many challenges. Finally, on the basis of precise positioning, a complete set of safety warning mechanisms needs to be constructed and implemented. The safety warning mechanism needs to comprehensively consider various factors, such as the location, behavior, and environment of personnel, and through intelligent algorithms and big data analysis, realize early warning and active prevention and control of potential safety hazards to ensure the lives of coal mine personnel.

[0004] Therefore, in order to overcome the defects of traditional technologies, improve positioning accuracy, enhance system stability, and provide efficient and timely safety warnings, a new method for positioning coal mine personnel and safety warning is studied and developed to improve the safety guarantee level of coal mine workers, promote the development of coal mine safety management towards intelligence and automation, and enhance the overall safety production capacity of the coal mine industry. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for positioning coal mine personnel and safety warning, which can realize precise positioning, real-time monitoring of the positions of coal mine personnel, and early warning of abnormal behaviors, and improve the level of coal mine safety management and accident prevention ability.

[0006] To achieve the above purpose, the present invention provides the following solutions:

[0007] A method for positioning coal mine personnel and safety warning includes:

[0008] Obtain the location information of coal mine personnel;

[0009] Process the location information to obtain high-precision positioning data;

[0010] Analyze the high-precision positioning data through a clustering algorithm to mark potential safety hazard data points;

[0011] Use the isolation forest algorithm to perform anomaly detection on the potential safety hazard data points to obtain abnormal behavior characteristics and potential safety hazards;

[0012] Based on the abnormal behavior characteristics and potential safety hazards, conduct a risk assessment to obtain a risk assessment value, compare the risk assessment value with a preset risk threshold, and trigger a safety warning according to the comparison result.

[0013] Optionally, obtaining the location information of the coal mine personnel includes: obtaining the UWB positioning data, RFID positioning data, and inertial navigation positioning data of the coal mine personnel, where the inertial navigation positioning data is used as compensation data for the UWB positioning data and RFID positioning data when the environmental interference intensity is large.

[0014] Optionally, processing the location information to obtain high-precision positioning data includes:

[0015] By analyzing the correlation between UWB positioning data, RFID positioning data, and inertial navigation positioning data, determine the weight coefficients of data fusion and construct a weighted data fusion model;

[0016] Input the UWB positioning data, RFID positioning data, and inertial navigation positioning data of the coal mine personnel into the weighted data fusion model to obtain an initial value of the fused high-precision positioning data;

[0017] Perform noise detection and noise removal on the initial value of the fused high-precision positioning data through an adaptive filtering algorithm to obtain high-precision positioning data.

[0018] Optionally, analyzing the high-precision positioning data through a clustering algorithm to mark potential safety hazard data points includes:

[0019] Preprocess and extract features from the high-precision positioning data to obtain a data set;

[0020] Use the clustering algorithm to perform clustering analysis on the data set, and divide similar data points into the same cluster by calculating the similarity between data points;

[0021] According to the clustering analysis result, judge whether there is abnormal behavior. If there is abnormal behavior, mark the corresponding data points as potential safety hazard data points.

[0022] Optionally, the Isolation Forest algorithm is used to perform anomaly detection on the potential safety hazard data points, and the obtained abnormal behavior characteristics and potential safety hazards include:

[0023] The Isolation Forest algorithm is used to calculate the anomaly score and abnormal behavior characteristics of the potential safety hazard data points, and the anomaly degree and abnormal behavior category are determined based on the anomaly score and abnormal behavior characteristics;

[0024] Through a preset abnormal behavior feature library, the association relationship and frequent pattern between each type of abnormal behavior and other abnormal behaviors are obtained, and the development law of each type of abnormal behavior, that is, the potential safety hazard, is obtained. Among them, the abnormal behavior feature library is constructed by using the association rule mining algorithm based on the behavior pattern and time series characteristics of each type of abnormal behavior.

[0025] Optionally, risk assessment is performed based on the abnormal behavior characteristics and potential safety hazards, and the obtained risk assessment value includes:

[0026] Collect the behavior data and environmental data of coal mine personnel;

[0027] After cleaning, feature extraction, data standardization of the behavior data and environmental data, and corresponding risk assessment value annotation, a training data set is constructed;

[0028] Based on the training data set, a neural network model is trained, and the model parameters are continuously optimized through the backpropagation algorithm to obtain a risk assessment model, where the neural network model includes a convolutional neural network and a long short-term memory network;

[0029] The abnormal behavior characteristics and potential safety hazards are input into the risk assessment model, and the risk assessment value is output.

[0030] Optionally, the method further includes: using the low-power wide-area network technology to transmit the high-precision positioning data. If the transmission signal decays below the preset signal threshold, a relay node is used for data forwarding.

[0031] Optionally, the method further includes: storing the high-precision positioning data by using a distributed storage architecture.

[0032] The beneficial effects of the present invention are:

[0033] The present invention obtains the initial position of personnel through multi-source fusion positioning technology, and uses inertial navigation to supplement data under environmental interference; adopts Kalman filtering and adaptive filtering algorithms to fuse and denoise multi-source data to obtain high-precision positioning data; transmits data through low-power wide-area network technology, and uses relay nodes to ensure real-time performance and reliability when necessary; adopts a distributed storage architecture and data compression algorithm to efficiently store positioning data. Analyze data using clustering algorithms and isolation forest algorithms to detect abnormal behavior patterns; finally, jointly model personnel behavior and environmental data through a neural network model, and trigger an early warning mechanism when the risk threshold is exceeded. The present invention realizes precise positioning, real-time monitoring and abnormal behavior early warning of coal mine personnel, and significantly improves the coal mine safety management level and accident prevention ability. Brief Description of the Drawings

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0035] Figure 1 It is a flow chart of a method for positioning and safety early warning of coal mine personnel in an embodiment of the present invention. Detailed Embodiments

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0037] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0038] This embodiment provides a method for positioning and safety early warning of coal mine personnel, as Figure 1 shown, including:

[0039] Obtain the position information of coal mine personnel;

[0040] Process the position information to obtain high-precision positioning data;

[0041] Analyze the high-precision positioning data through a clustering algorithm to mark potential safety hazard data points;

[0042] The isolated forest algorithm is used to detect anomalies in potential safety hazard data points, obtaining abnormal behavior characteristics and potential safety hazards;

[0043] Based on the abnormal behavior characteristics and potential safety hazards, risk assessment is carried out to obtain a risk assessment value, and the risk assessment value is compared with a preset risk threshold, and a safety warning is triggered according to the comparison result.

[0044] Specifically, in this embodiment, the initial position of the personnel is obtained through multi-source fusion positioning technology, and inertial navigation is used to supplement data under environmental interference; the Kalman filter and adaptive filter algorithms are used to fuse and denoise multi-source data to obtain high-precision positioning data; the data is transmitted through low-power wide-area network technology, and relay nodes are used when necessary to ensure real-time performance and reliability; a distributed storage architecture and data compression algorithm are used to efficiently store positioning data. The clustering algorithm and isolated forest algorithm are used to analyze the data to detect abnormal behavior patterns; finally, a neural network model is used to jointly model the personnel behavior and environmental data, and a warning mechanism is triggered when the risk threshold is exceeded. The precise positioning, real-time monitoring and abnormal behavior warning of coal mine personnel are realized, significantly improving the coal mine safety management level and accident prevention ability.

[0045] Furthermore, obtaining the position information of coal mine personnel includes: obtaining UWB positioning data, RFID positioning data and inertial navigation positioning data of coal mine personnel, where the inertial navigation positioning data is used as compensation data for UWB positioning data and RFID positioning data when the environmental interference intensity is large.

[0046] Specifically, this embodiment obtains the initial position information of coal mine personnel and uses multi-source fusion positioning technologies such as UWB positioning data, RFID positioning data and inertial navigation for positioning. According to the environmental interference intensity, it is judged whether supplementary positioning data is needed. If the environmental interference is strong, supplementary positioning data is obtained through inertial navigation and fused with UWB positioning data and RFID positioning data to obtain a preliminary positioning result.

[0047] Coal mine personnel positioning adopts multi-source fusion technology, combining UWB positioning data, RFID positioning data and inertial navigation and other methods to obtain initial position information. UWB technology uses ultra-wideband signals to calculate the distance by measuring the signal propagation time, which is suitable for high-precision positioning in complex environments. RFID technology identifies tags through electromagnetic field induction and can achieve fast positioning in narrow spaces. The inertial navigation system measures the motion state using accelerometers and gyroscopes and is suitable for dynamic tracking. In practical applications, the positioning strategy is dynamically adjusted according to the environmental interference intensity. For example, when there are a large number of metal devices in the coal mine roadway, it may interfere with the propagation of UWB signals. At this time, the weight of inertial navigation will be increased, and the positioning accuracy will be improved through the fusion algorithm. The fused data is processed by Kalman filtering, which can effectively eliminate random errors and make the positioning result smoother and more stable.

[0048] Furthermore, processing the position information to obtain high-precision positioning data includes:

[0049] By analyzing the correlation between UWB positioning data, RFID positioning data and inertial navigation positioning data, determine the weight coefficients of data fusion and construct a weighted data fusion model;

[0050] Input the UWB positioning data, RFID positioning data and inertial navigation positioning data of coal mine personnel into the weighted data fusion model to obtain the initial value of the fused high-precision positioning data;

[0051] Perform noise detection and noise removal on the initial value of the fused high-precision positioning data through an adaptive filtering algorithm to obtain high-precision positioning data.

[0052] Specifically, in this embodiment, according to the preliminary positioning result, multi-source heterogeneous sensor data is obtained. For different types of data sources, corresponding data preprocessing methods are used for standardization. Through correlation analysis, the correlation degree between multi-source data is judged, the weight coefficients of data fusion are determined, and a weighted data fusion model is constructed. The Kalman filtering algorithm is used to input multi-source data into the fusion model for processing to obtain the initial value of the fused high-precision positioning data. Noise detection is performed on the fused positioning data. If the data is interfered by noise, the filtering algorithm is adaptively selected according to the noise characteristics for denoising. The noisy data is processed through an adaptive filtering algorithm, the filter parameters are automatically adjusted, the noise is effectively suppressed, and the filtered high-precision positioning data is obtained. The filtered positioning data is compared and corrected with the prior position information, and the positioning deviation is corrected through error compensation to further improve the positioning accuracy. According to the application scenario requirements, the parameters of the Kalman filter and the adaptive filter are dynamically adjusted to optimize the fusion positioning algorithm model to ensure the continuity and stability of the positioning data.

[0053] The acquisition of multi-source heterogeneous sensor data is the key to improving positioning accuracy. Taking the coal mine environment as an example, multiple sensors such as UWB, RFID, and inertial navigation can be used. UWB technology has the characteristics of high precision and strong anti-multipath ability, and is suitable for complex underground environments; RFID technology has low cost and is easy to deploy, and can be used as an auxiliary positioning means; inertial navigation can provide continuous positioning when the signal is blocked. Data preprocessing is the basis for fusion. For UWB data, the time difference positioning algorithm can be used to calculate the initial position; for RFID data, the distance needs to be estimated according to the signal strength; for inertial navigation data, integral operations are required to obtain the displacement. By standardizing different data sources to a unified coordinate system, it lays the foundation for subsequent fusion. Correlation analysis helps to determine reasonable fusion weights. For example, when the UWB signal is good in an open area, a higher weight can be assigned to it; in a narrow roadway, the weights of RFID and inertial navigation can be appropriately increased. By calculating the Pearson correlation coefficient, the correlation degree between different data sources can be quantified, so as to dynamically adjust the fusion weights. Kalman filtering is an effective tool for realizing multi-source data fusion. It can optimally estimate the system state based on historical data and current observations. In coal mine positioning, the personnel position and speed can be used as state variables, and sensor data as observations. Through two stages of prediction and update, the positioning result can be continuously optimized. Noise detection and denoising are crucial for improving positioning accuracy. Abnormal noise can be identified by analyzing statistical features such as the variance and kurtosis of the data. For Gaussian white noise, Wiener filtering can be used; for burst noise, wavelet transform has better denoising effect. Adaptive filtering algorithms can automatically adjust parameters according to the noise characteristics. For example, the variable step size LMS algorithm can quickly converge when the signal changes suddenly. The introduction of prior position information helps to further improve the accuracy. For example, using the coal mine roadway layout map, the positioning result can be constrained within the effective area. Through error compensation techniques such as the least squares method, the system deviation can be effectively eliminated, making the positioning result closer to the actual situation. Dynamic parameter adjustment can adapt to the complex and changeable coal mine environment. For example, when the personnel are moving at high speed, the process noise covariance of Kalman filtering can be increased to improve the tracking ability of dynamic changes; in the stationary state, the measurement noise covariance can be reduced to improve the steady-state accuracy. By real-time monitoring the positioning error, the parameter adaptive adjustment mechanism can be triggered to ensure that the algorithm performance is always in the optimal state.

[0054] Obtain the original positioning data containing noise interference. According to the characteristics of the positioning data, select a suitable adaptive filtering algorithm for processing. According to the adaptive filtering algorithm, automatically adjust the parameter settings of the filter, including the order of the filter, cut-off frequency, etc., to adapt to the positioning data with different noise characteristics. Input the original positioning data into the adaptive filter with adjusted parameters for filtering. The adaptive filter automatically adjusts the filtering coefficients according to the noise characteristics to effectively suppress the noise. During the filtering process, the adaptive filter continuously monitors the change in the signal-to-noise ratio of the filtered output data. If the signal-to-noise ratio is lower than the preset threshold, it will trigger the automatic adjustment of the filter parameters again until the required signal-to-noise ratio requirement is met. After filtering, high-precision positioning data with suppressed noise interference is obtained. Compared with the original noisy data, this positioning data has a higher signal-to-noise ratio and positioning accuracy. To further improve the filtering performance, the Kalman filtering algorithm can be combined to fuse the positioning data after adaptive filtering. The optimal value of the positioning data is estimated through Kalman filtering to obtain a smoother and more accurate positioning result. Combining the processing results of adaptive filtering and Kalman filtering, the final high-precision positioning data is output. This positioning data not only suppresses noise interference but also has higher smoothness and continuity, meeting the application requirements of high-precision positioning.

[0055] The adaptive filtering algorithm is a technique that can automatically adjust the filter parameters according to the characteristics of the input signal. In the processing of positioning data, this algorithm is particularly important because the noise characteristics may vary significantly in different environments. To further improve the positioning accuracy, adaptive filtering can be combined with Kalman filtering. Kalman filtering can use the dynamic model of the system and the statistical characteristics of the measurement values to perform optimal estimation on the positioning data. The advantage of this combined method is that adaptive filtering can effectively handle instantaneous noise, while Kalman filtering can provide long-term trajectory smoothing and prediction. The final output high-precision positioning data not only has effectively suppressed noise but also has better continuity and predictability. This is of great significance for applications such as real-time navigation and trajectory tracking, which can significantly improve the reliability of the system and the user experience. Through this multi-level filtering and fusion processing, the original positioning data that may be full of noise and uncertainty can be transformed into high-quality position information, providing a reliable basis for various applications that rely on precise positioning. The flexibility and adaptability of this method enable it to remain efficient in different environments and conditions, thus meeting the increasing accuracy requirements of modern positioning systems.

[0056] Furthermore, analyze the high-precision positioning data through a clustering algorithm and mark the potential safety hazard data points, including:

[0057] Preprocess and extract features from the high-precision positioning data to obtain a data set;

[0058] The clustering algorithm is used to perform clustering analysis on the data set. By calculating the similarity between data points, similar data points are divided into the same cluster;

[0059] According to the results of the clustering analysis, it is judged whether there is abnormal behavior. If there is abnormal behavior, the corresponding data points are marked as potential safety hazard data points.

[0060] Specifically, in this embodiment, data preprocessing and feature extraction are performed on the high-precision positioning data to obtain a data set suitable for clustering analysis. According to the preset clustering algorithm, clustering analysis is performed on the data set. By calculating the similarity between data points, similar data points are divided into the same cluster. For the results of the clustering analysis, it is judged whether there is an abnormal behavior pattern. If there is an abnormal behavior pattern, the corresponding data points are marked as potential safety hazards.

[0061] After the high-precision positioning data is preprocessed and feature-extracted, a data set suitable for clustering analysis is formed. This process may include operations such as data cleaning, normalization, and dimensionality reduction. Clustering analysis is the process of dividing similar data points into the same category. Commonly used clustering algorithms include K-means, DBSCAN, etc. Taking the trajectory data of coal miners as an example, clustering can be performed according to the walking routes and stopping points of coal miners to identify regular walking routes and regular stopping areas. In the clustering results, some abnormal behavior patterns may be found. For example, some coal miners frequently appear in unconventional stopping areas, or the walking routes are significantly different from the regular walking routes. These data points will be marked as potential safety hazard data points.

[0062] Furthermore, the isolation forest algorithm is used to perform anomaly detection on the potential safety hazard data points, and the abnormal behavior characteristics and potential safety hazards obtained include:

[0063] The isolation forest algorithm is used to calculate the anomaly scores and abnormal behavior characteristics of the potential safety hazard data points, and determine the anomaly degree and anomaly behavior category based on the anomaly scores and abnormal behavior characteristics;

[0064] Through the preset abnormal behavior feature library, the association relationships and frequent patterns between each type of abnormal behavior and other abnormal behaviors are obtained, and the development rules of each type of abnormal behavior, that is, potential safety hazards, are obtained. Among them, the abnormal behavior feature library is constructed by using the association rule mining algorithm based on the behavior patterns and time series characteristics of each type of abnormal behavior.

[0065] Specifically, this embodiment uses the isolation forest algorithm to perform anomaly detection on the data points marked as potential safety hazards. By constructing isolation trees and calculating anomaly scores, the anomaly degree of the data points is determined. According to the anomaly detection results of the isolation forest algorithm, further analysis is performed on the abnormal data points to obtain the specific characteristics and potential risks of the abnormal behavior.

[0066] Based on the anomaly detection results of the Isolation Forest algorithm, relevant information of abnormal data points is obtained, including eigenvalue, anomaly score, etc. of the data points. The eigenvalues of the abnormal data points are analyzed in depth, and the abnormal data points are grouped according to feature similarity through a clustering algorithm to identify different types of abnormal behaviors. For each type of abnormal behavior, its specific features are extracted, including behavior patterns, time series features, etc., to construct an abnormal behavior feature library. According to the abnormal behavior feature library, an association rule mining algorithm is used to discover the association relationships and frequent patterns among different abnormal behaviors, and to reveal the potential laws of abnormal behaviors.

[0067] Furthermore, a risk assessment is carried out based on abnormal behavior features and potential safety hazards, and the obtained risk assessment values include:

[0068] Collect the behavior data and environmental data of coal mine workers;

[0069] After cleaning, feature extraction, data standardization of the behavior data and environmental data, and annotation of the corresponding risk assessment values, a training data set is constructed;

[0070] Based on the training data set, a neural network model is trained, and the model parameters are continuously optimized through the backpropagation algorithm to obtain a risk assessment model, where the neural network model includes a convolutional neural network and a long short-term memory network;

[0071] Input the abnormal behavior features and potential safety hazards into the risk assessment model, and output the risk assessment value.

[0072] Specifically, the behavior data and environmental data of personnel are collected through Internet of Things devices, including personnel movement trajectories, equipment status, environmental parameters, etc., and the data is transmitted to the cloud platform for storage and processing. The collected data is preprocessed, including data cleaning, feature extraction, data standardization, etc., and the corresponding risk assessment values are annotated for the preprocessed behavior data and environmental data to prepare for subsequent model training and analysis. According to business requirements and data characteristics, a suitable neural network model is selected, such as a convolutional neural network (CNN), a long short-term memory network (LSTM), etc., to jointly model the personnel behavior and environmental data. The neural network model is trained using the training set, and the model parameters are continuously optimized through the backpropagation algorithm to improve the prediction accuracy and generalization ability of the model. After the model training is completed, the abnormal behavior features and potential safety hazards obtained previously are input into the model to obtain the output result of the model, that is, the risk assessment value of the potential safety hazard.

[0073] The selection of a neural network model needs to consider the data characteristics and task requirements. For the behavioral sequence data of coal miners, the long short-term memory network (LSTM) can effectively capture the temporal dependence relationship. For example, LSTM can be used to predict whether a worker is likely to enter a dangerous area. The convolutional neural network (CNN), on the other hand, is suitable for processing image data, such as identifying whether a coal miner is wearing a safety helmet correctly. During the model training process, historical accident data can be used as labels. For example, the behavioral sequences of coal miners who have experienced safety accidents in the past are marked as high-risk samples. Through the backpropagation algorithm, the model can learn the potential risk patterns. To improve the generalization ability of the model, data augmentation techniques can be adopted, such as adding random noise to simulate sensor errors. Real-time risk assessment is the core of model application. When new data is input into the model, the output risk assessment value may range from 0 to 1. Suppose the risk threshold is set at 0.8. When the behavioral sequence of a certain coal miner causes the model to output a risk value of 0.9, an early warning will be immediately triggered. The early warning information is automatically notified to the corresponding personnel through the underground IP broadcast and sent to the safety supervisor via text message. At the same time, the emergency response plan is activated, such as automatically shutting down the dangerous equipment in the relevant area. Continuous optimization is the key to maintaining the effectiveness of the model. As new equipment is introduced or the production process is adjusted in the coal mine, the original risk patterns may change. Regularly retraining the model can adapt to these changes. For example, use the data of the latest month to fine-tune the model every month to ensure that the model can identify the newly emerging risk patterns. In this way, the intelligent safety management system can continuously improve the accuracy of risk prediction and provide strong guarantee for the safe production of coal mines.

[0074] Furthermore, the method provided in this embodiment further includes: using the low-power wide-area network technology to transmit high-precision positioning data. If the transmission signal decays below the preset signal threshold, a relay node is used for data forwarding.

[0075] Specifically, for the positioning accuracy and data volume of high-precision positioning data, select an appropriate data compression algorithm to reduce the data transmission volume while ensuring the positioning accuracy. According to the network coverage and transmission distance, dynamically adjust the transmit power and transmission rate to balance power consumption and transmission efficiency in a low-power wide-area network. Through machine learning algorithms, based on historical data, predict the signal attenuation trend. When it is judged that a serious signal attenuation may occur, start the relay node in advance to prepare for data forwarding. Real-time monitor the network status and data transmission quality. When the real-time performance and reliability of data transmission drop to a preset threshold, dynamically optimize the routing and select the optimal relay node for data forwarding. Adopt an adaptive retransmission mechanism to dynamically adjust the retransmission times and time intervals according to the transmission distance and signal quality, reduce unnecessary retransmissions while ensuring transmission reliability, and reduce energy consumption. For different types of positioning data and service requirements, adopt a differentiated transmission strategy. Give priority to transmitting data with high real-time requirements; increase the transmission retry times for data with high accuracy requirements. Through machine learning algorithms, continuously optimize the network parameters and transmission strategy, and dynamically adjust the system configuration according to the changes in the network status of the low-power wide-area network to achieve a dynamic balance between network performance and energy efficiency.

[0076] Specifically, to achieve efficient transmission of high-precision positioning data in a low-power wide-area network, multiple aspects such as data compression, power consumption management, signal prediction, and routing optimization need to be comprehensively considered. First, for high-precision positioning data, differential compression algorithms can be used for processing. For example, for GPS positioning data, only the difference between two adjacent sampling points can be transmitted instead of the complete coordinate information, which can significantly reduce the data volume. In practical applications, if the original data accuracy is centimeter-level, the data volume can be reduced by more than 50% after compression while maintaining the positioning accuracy unchanged. To balance power consumption and transmission efficiency, the system can dynamically adjust the transmission power according to the network coverage. In areas with good signal, the transmission power can be reduced to 10 mW, while in areas with poor signal, it can be increased to 100 mW. At the same time, the transmission rate can also be adjusted according to the distance. Higher rates, such as 50 kbps, can be used for short-distance transmission, while it can be reduced to 10 kbps for long-distance transmission to ensure transmission reliability. By using machine learning algorithms to predict the signal attenuation trend, countermeasures can be taken in advance. For example, using the Long Short-Term Memory (LSTM) model to predict the signal strength change within the next hour based on historical data. When it is predicted that the signal strength may drop below -100 dBm, the system will activate nearby relay nodes in advance to ensure the continuity of data transmission. Real-time monitoring of the network status is crucial for maintaining the transmission quality. The system can set the data transmission delay threshold to 200 ms and the packet loss rate threshold to 1%. When these metrics are monitored to exceed the thresholds, the routing optimization algorithm, such as the AODV (Adhoc On-Demand Distance Vector) protocol, will be immediately triggered to recalculate the optimal path and select relay nodes with better signal quality for data forwarding. The adaptive retransmission mechanism can be dynamically adjusted according to the transmission environment. For example, in the case of short-distance transmission (such as within 1 km) and good signal quality (RSSI > -90 dBm), the retransmission times can be set to 2 times and the retransmission interval to 100 ms. While in the case of long-distance transmission (such as more than 5 km) or poor signal quality (RSSI < -100 dBm), the retransmission times can be increased to 5 times and the retransmission interval extended to 500 ms to improve the transmission success rate. For different types of positioning data, differentiated transmission strategies can be adopted. For data with high real-time requirements, the highest priority can be given to ensure transmission within 50 ms. While for data with high accuracy requirements but low real-time requirements, the transmission retry times can be increased to 10 times to ensure the integrity and accuracy of the data. Through continuous learning and optimization, the system can continuously improve its performance. For example, using reinforcement learning algorithms to automatically adjust transmission parameters according to metrics such as network throughput, latency, and energy consumption. In practice, this adaptive optimization can improve the overall network performance by more than 20% while reducing energy consumption by 30%. This dynamic balance strategy ensures that the low-power wide-area network always maintains an efficient and stable operating state in a complex and changing environment.

[0077] Further, the method provided in this embodiment further includes: storing the high-precision positioning data using a distributed storage architecture.

[0078] Specifically, according to the configuration information of the distributed storage architecture, the high-precision positioning data is distributed to different storage nodes for storage, which can improve the storage capacity and parallel processing ability. For example, in this embodiment, the positioning system in a certain coal mine generates hundreds of GB of positioning data every day. By dispersing the data and storing it on multiple nodes, it can not only avoid single-point failures but also improve the data reading and writing efficiency.

[0079] Count the amount of positioning data on each storage node and summarize to obtain the total amount of positioning data of the entire system. Compare the total amount of positioning data with a preset threshold, and perform dynamic expansion based on the comparison result. When the load exceeds the preset threshold, resources are automatically increased (vertical or horizontal expansion); conversely, when the load decreases, resources are automatically reduced. In this embodiment, the dynamic expansion is performed through the auto-scaling service provided by a cloud computing platform (such as AWS, Azure, Google Cloud, etc.).

[0080] At the same time, regularly perform consistency checks on the stored data of each storage node to ensure that there are no inconsistencies in the content, status, etc. of the data copies on different nodes in the distributed storage environment, and avoid data loss, errors, or service exceptions caused by data inconsistencies. Specifically, a consistent hashing algorithm can be used to detect whether the data is consistent by using hash values, ensuring that the data on each node in the storage system can be consistent among different nodes.

[0081] As another alternative embodiment, data compression can also be triggered by the comparison result. When the load exceeds the threshold, data compression is triggered; otherwise, the original storage method is maintained. For the storage nodes that need to be compressed, a data compression algorithm is used to compress the positioning data to reduce the storage space occupied by the data. Compression algorithms that can be selected include run-length encoding, dictionary encoding, etc. The compressed positioning data is re-stored to the corresponding storage node, and the data volume statistics information of the node is updated. According to the data distribution of each storage node, a load balancing algorithm is used to dynamically adjust the distribution of data among different nodes to avoid excessive data volume on a single node. An index is established for the compressed positioning data to improve the data retrieval and access efficiency, ensuring that the compressed data can still meet the real-time requirements of the business.

[0082] Among them, run-length encoding is applicable to data with many consecutive repeated values, such as the positioning information of stationary objects. Dictionary encoding is more suitable for processing data with certain regularity, such as the walking trajectories on fixed routes. By reasonably selecting the compression algorithm, the storage space occupancy can be significantly reduced while ensuring the data quality. Re-storing the compressed data and updating the statistical information are necessary steps to maintain storage consistency. This not only involves the physical storage of data but also includes the update of metadata. For example, a storage node originally stored 1TB of data, and after compression, it may only occupy 300GB. The system needs to update this information in a timely manner for subsequent load balancing and resource scheduling. The load balancing algorithm plays an important role in a distributed system. Common algorithms include the round-robin method, the least-connections method, etc. Taking the round-robin method as an example, the system can sequentially allocate the newly added positioning data to different storage nodes to ensure uniform data distribution. This method is simple and effective, but it may not be able to handle the situation where the node performance differences are large. Therefore, in practical applications, it is often necessary to dynamically adjust the data distribution strategy by combining factors such as the processing capacity and storage capacity of the nodes. Establishing an index for the compressed positioning data is the key to improving the data retrieval efficiency. Common index technologies include B-trees, hash indexes, etc. For example, a multi-dimensional index can be established based on the timestamp and geographical location information, enabling the system to quickly respond to complex queries such as "finding all the positioning data in a certain coal mine area during a certain period". An efficient index can not only improve the data access speed but also reduce the system resource consumption, providing strong support for real-time services. Through the comprehensive application of these technologies, the high-precision positioning system can efficiently manage a large amount of positioning data with limited storage resources and provide reliable data support for various location services.

[0083] The embodiments described above are only descriptions of the preferred embodiments of the present invention and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for positioning and safety warning of coal mine personnel, characterized in that Including: Obtaining the location information of coal mine personnel; Processing the location information to obtain high-precision positioning data; Analyzing the high-precision positioning data through a clustering algorithm to mark potential safety hazard data points; Using an isolation forest algorithm to perform anomaly detection on the potential safety hazard data points to obtain anomaly behavior characteristics and potential safety hazards; Based on the anomaly behavior characteristics and potential safety hazards, performing risk assessment to obtain a risk assessment value, comparing the risk assessment value with a preset risk threshold, and triggering a safety warning according to the comparison result.

2. The coal mine personnel positioning and safety warning method according to claim 1, wherein Obtaining the location information of the coal mine personnel includes: obtaining the UWB positioning data, RFID positioning data, and inertial navigation positioning data of the coal mine personnel, where the inertial navigation positioning data is used as compensation data for the UWB positioning data and RFID positioning data when the environmental interference intensity is high.

3. The coal mine personnel positioning and safety warning method according to claim 2, characterized in that, Processing the location information to obtain high-precision positioning data includes: By analyzing the correlation between the UWB positioning data, RFID positioning data, and inertial navigation positioning data, determining the weight coefficients of data fusion and constructing a weighted data fusion model; Inputting the UWB positioning data, RFID positioning data, and inertial navigation positioning data of the coal mine personnel into the weighted data fusion model to obtain an initial value of the fused high-precision positioning data; Performing noise detection and noise removal on the initial value of the fused high-precision positioning data through an adaptive filtering algorithm to obtain high-precision positioning data.

4. The coal mine personnel positioning and safety warning method according to claim 1, wherein Analyzing the high-precision positioning data through a clustering algorithm to mark potential safety hazard data points includes: Preprocessing and feature extraction of the high-precision positioning data to obtain a data set; Using the clustering algorithm to perform clustering analysis on the data set, and by calculating the similarity between data points, dividing similar data points into the same cluster; According to the clustering analysis result, determining whether there is abnormal behavior. If there is abnormal behavior, the corresponding data points are marked as potential safety hazard data points.

5. The coal mine personnel positioning and safety warning method according to claim 1, characterized in that Using an isolation forest algorithm to perform anomaly detection on the potential safety hazard data points to obtain anomaly behavior characteristics and potential safety hazards includes: Using an isolation forest algorithm to calculate the anomaly score and anomaly behavior characteristics of the potential safety hazard data points, and determining the anomaly degree and anomaly behavior category based on the anomaly score and anomaly behavior characteristics; Through a preset anomaly behavior feature library, obtaining the correlation relationship and frequent pattern between each type of anomaly behavior and other anomaly behaviors, and obtaining the development law of each type of anomaly behavior, that is, the potential safety hazard, where the anomaly behavior feature library is constructed based on the behavior pattern and time series characteristics of each type of anomaly behavior using an association rule mining algorithm.

6. The coal mine personnel positioning and safety warning method according to claim 1, characterized in that, Based on the anomaly behavior characteristics and potential safety hazards, performing risk assessment to obtain a risk assessment value includes: Collecting the behavior data and environmental data of coal mine personnel; After performing data cleaning, feature extraction, data standardization, and marking the corresponding risk assessment values on the behavior data and environmental data, constructing a training data set; Train a neural network model based on the training data set, and continuously optimize the model parameters through the backpropagation algorithm to obtain a risk assessment model, where the neural network model includes a convolutional neural network and a long short-term memory network; Input the abnormal behavior features and potential safety hazards into the risk assessment model, and output the risk assessment value.

7. The coal mine personnel positioning and safety warning method according to claim 1, characterized in that The method further includes: transmitting the high-precision positioning data by using a low-power wide-area network technology, and if the transmission signal decays below a preset signal threshold, using a relay node for data forwarding.

8. The coal mine personnel positioning and safety warning method according to claim 1, characterized in that, The method further includes: storing the high-precision positioning data by using a distributed storage architecture.

Images