Mine audio communication control method and system

By analyzing the mine environment through distributed sensor networks and machine learning algorithms, and dynamically adjusting communication paths and encoding strategies, the problem of unstable underground communication in mines was solved, and stable and reliable mine audio communication was achieved.

CN118694452BActive Publication Date: 2024-11-08YANKUANG ENERGY GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411180155.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-11-08
Estimated Expiration
2044-08-27

AI Technical Summary

Technical Problem

In the complex underground environment, underground audio communication systems in mines suffer from unstable communication and degraded signal quality. Existing technologies lack flexibility and adaptability, making it difficult to accurately predict and adapt to environmental changes.

Method used

A distributed sensor network is used to monitor the electrical conductivity and gas concentration of rock strata. An improved LSTM model is used to analyze electromagnetic wave propagation conditions. The transmission path is dynamically adjusted by combining the Dijkstra algorithm and the coding strategy is optimized by Bayesian network to achieve adaptive communication.

Benefits of technology

It improves the stability and reliability of mine communication, dynamically avoids interference and attenuation areas, optimizes electromagnetic wave transmission, and reduces communication interruptions and data loss.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118694452B_ABST
    Figure CN118694452B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of coal mine communication technology, and specifically to a method and system based on mine audio communication control, including the following steps: implementing real-time monitoring of rock stratum conductivity, using a sensor network distributed in key areas of coal mines, and continuously monitoring changes in rock stratum conductivity; configuring coal seam gas monitoring equipment to detect gas concentrations in each operating area in real time; in a central processing unit, according to the received rock stratum conductivity and gas concentration distribution data, using a machine learning algorithm to analyze the electromagnetic wave propagation conditions of the current underground environment, and predict the quality and stability of the communication link; according to the electromagnetic wave propagation conditions and the communication link quality evaluation, dynamically adjusting the transmission path; and implementing environmental adaptive signal coding technology. The present invention enhances the adaptive ability of wireless electromagnetic wave communication and maximizes the efficiency and reliability of communication.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mine communication, and in particular to a mine audio communication control method and system. Background Art

[0002] In the underground audio communication system of mines, stable and reliable communication is the key to ensure the safety of miners and improve production efficiency. Traditional underground audio communication systems in coal mines face multiple challenges, including complex underground environments, uneven rock conductivity, and the presence of gases such as gas, which may interfere with the propagation of electromagnetic waves, resulting in unstable communication and degraded signal quality. In addition, traditional methods generally lack flexibility and adaptability in dynamically adjusting communication links to cope with environmental changes.

[0003] In response to these problems, in recent years, some studies have attempted to use advanced communication technologies and machine learning algorithms to optimize mine communication networks. These methods aim to improve the robustness and reliability of communication systems by real-time monitoring of underground environmental parameters (such as rock conductivity and gas concentration) and dynamically adjusting communication links. However, existing technologies still face the challenge of how to accurately predict and adapt to complex environmental changes, and on this basis, achieve the need for efficient and stable communication.

[0004] Therefore, it is particularly important to develop a system that can accurately analyze the impact of the underground environment, predict the quality of the communication link, and dynamically adjust the communication strategy based on real-time environmental data, so as to maintain stable and reliable communication in the ever-changing mine environment. Summary of the invention

[0005] Based on the above objectives, the present invention provides a mine audio communication control method and system.

[0006] The present invention is based on a mine audio communication control method and comprises the following steps:

[0007] S1: Implement real-time monitoring of rock conductivity, use a sensor network distributed in key areas of coal mines, continuously monitor changes in rock conductivity, and transmit rock conductivity data to a central processing unit in real time;

[0008] S2: Equip coal seam gas monitoring equipment to detect the gas concentration in each operating area in real time and transmit the gas concentration distribution data to the central processing unit;

[0009] S3: In the central processing unit, based on the received rock conductivity and gas concentration distribution data, the machine learning algorithm is used to analyze the electromagnetic wave propagation conditions of the current underground environment and predict the quality and stability of the communication link;

[0010] S4: Implement dynamic routing protocols, dynamically adjust the transmission path according to electromagnetic wave propagation conditions and communication link quality assessment, avoid areas with severe interference and attenuation, and deploy mobile wireless relay nodes. The wireless relay nodes automatically adjust their physical positions and configurations according to the dynamically adjusted transmission paths to build the optimal communication links.

[0011] S5: Implement environmental adaptive signal coding technology, dynamically adjust the coding scheme according to the real-time monitoring data of rock conductivity and gas concentration distribution in the optimal communication link, and use the optimization model to automatically optimize the redundancy and error correction capability of electromagnetic waves to adapt to the current transmission conditions.

[0012] Furthermore, S1 includes deploying multi-point conductivity sensors in the main passages and working surfaces underground in the coal mine to directly measure the conductivity of the rock, and the monitored conductivity data is transmitted to the central processing unit in real time via a wireless network.

[0013] Furthermore, the machine learning algorithm in S3 is based on an improved long short-term memory network LSTM model. The central processing unit first pre-processes the received formation conductivity and gas concentration data, uses the LSTM model to analyze the impact of formation conductivity and gas concentration on electromagnetic wave propagation, processes time series data, identifies patterns and trends in the data, and predicts the propagation performance of electromagnetic waves under different downhole environmental conditions;

[0014] Simulate and analyze communication links to predict link quality and stability under current mine conditions, including signal strength, signal-to-noise ratio, data transfer rate, and interruptions.

[0015] Furthermore, the use of the LSTM model to analyze the impact of rock conductivity and gas concentration on electromagnetic wave propagation specifically includes:

[0016] S31: Feature Engineering: Integrate the multi-dimensional features of rock conductivity, gas concentration distribution, and depth to construct a comprehensive feature set to capture the complex factors that affect the propagation of electromagnetic waves, perform spatiotemporal feature mapping, and integrate spatial position (such as depth from the entrance, specific tunnel location) and time factors into the comprehensive feature set to simulate the actual propagation process of electromagnetic waves in the mine;

[0017] S32: Model establishment and training: Build an LSTM model and design an LSTM network architecture, including multiple LSTM units, and the LSTM unit includes an input layer, multiple LSTM layers, and an output layer. The number and configuration of LSTM layers are determined according to the complexity of the problem and the amount of data. Use historical data to train the LSTM model. During training, learn how to predict the propagation performance of electromagnetic waves based on the historical and current values ​​of rock conductivity and gas concentration.

[0018] S33: Real-time analysis and prediction: Input data collected in real time The trained LSTM model is input. The model predicts the propagation performance of electromagnetic waves based on current and past data trends, including electromagnetic wave attenuation, signal attenuation, and interference areas. The output of the LSTM model is interpreted to determine the predicted value of the electromagnetic wave propagation performance and is used to dynamically adjust the communication strategy.

[0019] Furthermore, the spatiotemporal feature map includes in the input data In the process of combining spatial and temporal characteristics, the rock conductivity and gas concentration are combined with their corresponding spatial position and time information to form a complex input feature vector to capture the temporal and spatial dependencies, including:

[0020] Spatial location encoding: Assign a unique geographic tag to each monitoring point in the mine, use one-hot encoding to convert the geographic tag into a numerical form, and use it as an input feature in the model;

[0021] Time factor integration: The timestamp of data collection is converted into a format that the model can understand. This is achieved by calculating the time interval from the start of monitoring, converting the time into a continuous value that reflects the duration from the start of monitoring to the present. The converted time value is combined with conductivity and gas concentration to form a time series data set for training the LSTM model.

[0022] Combine spatiotemporal features for model training: Combine spatial position coding and time factors with sensor data to form a comprehensive feature vector. For the data at each time point, the feature vector can be [spatial coding, time value, conductivity, gas concentration]. Use the constructed comprehensive feature vector as the input of the LSTM model. The model also learns the spatiotemporal dependency in the process of electromagnetic wave propagation.

[0023] Furthermore, the S4 specifically includes:

[0024] Receiving LSTM model output: The central processing unit regularly receives the prediction data generated by the improved LSTM model, including the prediction of electromagnetic wave propagation conditions for each monitoring point;

[0025] Calculate the optimal path: Use the dynamic routing algorithm and the prediction results of the LSTM model to calculate the optimal communication path that avoids interference and attenuation areas;

[0026] Dynamically adjust the transmission path: According to the calculated optimal path, dynamically update the routing table and adjust the data transmission path to ensure that the electromagnetic waves are sent through the best path;

[0027] Deploy and adjust wireless relay nodes: According to the dynamically adjusted transmission path, guide the deployment and position adjustment of wireless relay nodes. The wireless relay nodes are equipped with automatic positioning and movement functions. According to the instructions of the central processing unit, adjust the physical position and configuration to support the new optimal communication link.

[0028] Furthermore, the dynamic routing algorithm uses a graph-based Dijkstra algorithm, specifically including:

[0029] Mine Propagation Network Modeling: Modeling electromagnetic wave propagation in mines as a weighted graph , where V is the node set in the graph, representing the wireless relay node, and E is the edge set, representing the communication link. There is a weight , represents the communication cost or quality from node u to node v. The weight reflects the interference level and attenuation factor of the signal. The calculation steps of the Dijkstra algorithm are as follows:

[0030] Initialization: For each node in the graph Assign a distance value , for the starting point ; For all other nodes, , create an empty set S to store the nodes whose shortest paths have been found;

[0031] Update distance: for nodes For each adjacent node n of , if n is not in S, update for ;

[0032] Determine the shortest path: Select the node with the smallest distance from the set of unprocessed nodes , add to set S, update node The distance value of the adjacent nodes;

[0033] Repeat updating the distance and determining the shortest path until all nodes are in S, that is, the shortest path from the starting point to all other nodes is found.

[0034] Furthermore, the optimization model in S5 adopts a Bayesian network model to process uncertainty and the probabilistic relationship between variables, selects electromagnetic wave coding strategies in a changing mine environment, constructs a Bayesian network model to represent the impact of rock conductivity and gas concentration on communication performance, and infers the optimal coding strategy.

[0035] Furthermore, the construction of the Bayesian network model specifically includes:

[0036] Define nodes: Create nodes to represent key variables that affect communication performance, including formation conductivity EDR, gas concentration GC, coding redundancy CR, and error correction level ECL; define the probabilistic dependency between key variables, that is, coding redundancy and error correction level depend on formation conductivity and gas concentration;

[0037] Parameter learning: Collect historical data, including formation conductivity, gas concentration, and the corresponding optimal coding redundancy and error correction level, and use the collected data to estimate the conditional probability distribution of each node in the Bayesian network;

[0038] Reasoning and decision-making: Real-time monitoring of the current rock conductivity and gas concentration. Given the current rock conductivity and gas concentration, Bayesian network is used for reasoning to calculate the posterior probability of different coding scheme parameters (CR and ECL);

[0039] Output recommendation plan: The posterior probability is calculated by Bayes' rule, and the formula is as follows:

[0040] ,in, is the posterior probability of the specific coding redundancy CR and error correction level ECL when the rock conductivity EDR and gas concentration GC are given, is the probability of observing a specific formation conductivity and gas concentration given the encoded parameters, learned from historical data, is the prior probability of coding redundancy and error correction level, is the marginal probability of the currently observed rock conductivity and gas concentration;

[0041] Select the best solution: Select the coding redundancy and error correction level that maximizes the posterior probability as the recommended solution.

[0042] A mine audio communication control system, used to implement the above-mentioned mine audio communication control method, includes the following modules:

[0043] Rock conductivity monitoring module: It uses a distributed sensor network, deployed in the mine area, to continuously monitor changes in rock conductivity and transmit data to the central processing unit in real time;

[0044] Gas concentration monitoring module: Use gas monitoring equipment, installed in the mine operation area, to detect gas concentration in real time and send distribution data to the central processing unit;

[0045] Data analysis and prediction module: includes a central processing unit equipped with machine learning algorithms to analyze rock conductivity and gas concentration data, predict electromagnetic wave propagation conditions and communication link quality and stability;

[0046] Dynamic routing adjustment module: It includes dynamic routing algorithms and mobile wireless relay nodes. It dynamically adjusts the transmission path according to electromagnetic wave propagation conditions and communication link quality evaluation, and automatically adjusts the location and configuration of wireless relay nodes.

[0047] Coding adjustment module: Based on environmental adaptive signal coding technology, the coding scheme is dynamically adjusted according to the real-time monitoring data in the communication link, and the redundancy and error correction capability of the electromagnetic wave are optimized to adapt to the transmission conditions.

[0048] Beneficial effects of the present invention:

[0049] The present invention can effectively predict the quality and stability of the communication link by using the improved LSTM model to analyze the electromagnetic wave propagation conditions of the current underground environment. It takes advantage of LSTM in processing time series data, so that the model can learn and identify the impact of changes in rock conductivity and gas concentration on electromagnetic wave propagation. Since the model can capture long-term data dependencies, it can predict future environmental changes and their impact on the performance of the communication link, so as to make adjustments in advance to maintain the continuity and stability of communication.

[0050] The present invention uses the graph-based Dijkstra algorithm to dynamically adjust the transmission path, which can effectively avoid areas with severe interference and attenuation. By treating the mine communication network as a weighted graph and updating the weight of each link in real time (reflecting the degree of interference and attenuation), the Dijkstra algorithm can calculate the optimal communication path, which not only enhances the adaptive ability of wireless electromagnetic wave communication, but also maximizes the efficiency and reliability of communication. In addition, the deployment of mobile wireless relay nodes further enhances the flexibility of the network, enabling the communication network to dynamically adapt to changes in the mine environment.

[0051] The present invention uses a Bayesian network model to handle the uncertainty of rock conductivity and gas concentration and their impact on communication performance. The Bayesian network formally describes the dependency between variables through a probabilistic method, and can infer the best electromagnetic wave coding strategy under the condition of constantly changing mine environment. This method enables the communication system to intelligently select coding redundancy and error correction level according to the current environmental conditions, optimize the data transmission process, and reduce the risk of communication interruption and data loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1A schematic diagram of a control method flow chart of an embodiment of the present invention;

[0054] Figure 2 Schematic diagram of control system functional modules according to an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0056] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0057] like Figure 1 As shown, based on the mine audio communication control method, the following steps are included:

[0058] S1: Implement real-time monitoring of rock conductivity, using a sensor network distributed in key areas of coal mines to continuously monitor changes in rock conductivity (uneven conditions), and transmit rock conductivity data to a central processing unit in real time;

[0059] S2: Equip coal seam gas monitoring equipment to detect the gas concentration in each operating area in real time and transmit the gas concentration distribution data to the central processing unit;

[0060] S3: In the central processing unit, based on the received rock conductivity and gas concentration distribution data, the machine learning algorithm is used to analyze the electromagnetic wave propagation conditions of the current underground environment and predict the quality and stability of the communication link;

[0061] S4: Implement dynamic routing protocols, dynamically adjust the transmission path according to electromagnetic wave propagation conditions and communication link quality assessment, avoid areas with severe interference and attenuation, and deploy mobile wireless relay nodes. The wireless relay nodes automatically adjust their physical positions and configurations according to the dynamically adjusted transmission paths to build the optimal communication links.

[0062] S5: Implement environmental adaptive signal coding technology. Since there are still slight uneven rock conductivity and slight gas concentration in the optimal communication link, the coding scheme is dynamically adjusted according to the real-time monitoring data of rock conductivity and gas concentration distribution in the optimal communication link. The optimization model is used to automatically optimize the redundancy and error correction capability of electromagnetic waves to adapt to the current transmission conditions, increase signal redundancy and adopt stronger error correction coding to resist the signal loss and interference that may increase during the transmission process.

[0063] S1 involves deploying multi-point conductivity sensors in the main passages and working surfaces underground in coal mines to directly measure the conductivity of rocks. Electrodes are usually used to measure the amount of current passing through rock samples to calculate the conductivity. The monitored conductivity data is transmitted to the central processing unit in real time via a wireless network.

[0064] The machine learning algorithm in S3 is based on an improved long short-term memory network (LSTM) model. The central processing unit first preprocesses the received formation conductivity and gas concentration data, and uses the LSTM model to analyze the impact of formation conductivity and gas concentration on electromagnetic wave propagation, process time series data, identify patterns and trends in the data, and predict the propagation performance of electromagnetic waves under different underground environmental conditions.

[0065] Simulate and analyze communication links to predict link quality and stability under current mine conditions, including signal strength, signal-to-noise ratio, data transfer rate, and interruptions.

[0066] The LSTM model is used to analyze the impact of rock conductivity and gas concentration on electromagnetic wave propagation, including:

[0067] S31: Feature Engineering: Integrate the multi-dimensional features of rock conductivity, gas concentration distribution, and depth to construct a comprehensive feature set to capture the complex factors that affect the propagation of electromagnetic waves, perform spatiotemporal feature mapping, and integrate spatial position (such as depth from the entrance, specific tunnel location) and time factors into the comprehensive feature set to simulate the actual propagation process of electromagnetic waves in the mine, such as converting time series data into sequence input;

[0068] S32: Model establishment and training: Build an LSTM model and design an LSTM network architecture, including multiple LSTM units, and the LSTM unit includes an input layer, multiple LSTM layers, and an output layer. The number and configuration of LSTM layers are determined according to the complexity of the problem and the amount of data. Use historical data to train the LSTM model. During training, learn how to predict the propagation performance of electromagnetic waves based on the historical and current values ​​of rock conductivity and gas concentration.

[0069] The LSTM unit consists of four parts: the forget gate , input gate , output gate and cell status , the expression is as follows:

[0070] Forget Gate: , controls how much previous information the memory unit should forget.

[0071] Input Gate: ;

[0072] ; Determines how much new information is stored in the cell state.

[0073] Unit status update: ; Update the unit state, which is the sum of the product of the state at the previous moment and the forget gate and the product of the input gate and the new candidate value.

[0074] Output Gate:

[0075] ;

[0076] ; Determine the output value of the unit at the next moment.

[0077] in, is the sigmoid activation function, which is used to limit the output of the forget gate, input gate, and output gate to between 0 and 1; tanh is the hyperbolic tangent activation function, which is used to limit the output range of the candidate unit state and the hidden state of the final output to between -1 and 1;

[0078] It means connecting the hidden state of the previous moment and the input of the current moment.

[0079] represents the hidden state at the previous moment, and represents the output state of the network at time step t-1.

[0080] represents the input at the current moment, and represents the external input received at time step t.

[0081] Represents the output of the forget gate, which is used to control the amount of information forgotten from the cell state.

[0082] The output of the input gate determines how much new information will be stored in the cell state.

[0083] It indicates the candidate unit state at the current moment and the new information content.

[0084] Represents the cell state at the current moment, updated by combining the previous state and the new candidate state.

[0085] Represents the output of the output gate, which is used to determine the amount of output from the cell state to the hidden state.

[0086] Represents the hidden state at the current moment, which is the final output determined by the output gate and the current unit state.

[0087] Represents the weight matrix of the forget gate, which is used to control the influence of the input and the previous hidden state on the output of the forget gate.

[0088] Represents the bias term of the forget gate, which is used to adjust the activation level of the forget gate.

[0089] Represents the weight matrix of the input gate, which is used to control the influence of the input and the previous hidden state on the output of the input gate.

[0090] Represents the bias term of the input gate, which is used to adjust the activation level of the input gate.

[0091] The weight matrix representing the candidate unit state is used to control the influence of the input and the previous hidden state on the candidate unit state.

[0092] The bias term representing the candidate unit state is used to adjust the activation level of the candidate unit state.

[0093] Represents the weight matrix of the output gate, which is used to control the influence of the input and the previous hidden state on the output gate.

[0094] Represents the bias term of the output gate, which is used to adjust the activation level of the output gate.

[0095] Further explanation in conjunction with the present invention is as follows:

[0096] 1. and Two variables are critical, as they directly process and integrate real-time monitoring data from the mine environment, including data collected from rock conductivity and gas concentration sensors, which are the basis for assessing the current underground environment and predicting the quality of the communication link.

[0097] 2. and The gating and state update mechanisms enable the LSTM to decide which old information to keep and when to introduce new information, which is critical for adjusting communication strategies under the dynamic and complex environmental conditions of the mine. For example, if the prediction model detects a significant change in the conductivity of the rock formation, the behavior of the forget gate and input gate needs to be adjusted to adapt to the new environmental conditions.

[0098] 3. Weight Matrix and Bias Term Parameters determine how LSTM learns from data and responds to environmental changes. In the mine communication control system, weights and biases are trained based on historical data so that the model can optimally predict the impact of different environmental variables such as rock conductivity and gas concentration on communication effects.

[0099] 4. Activation function ( The choice of activation function is crucial to ensuring the stability and effectiveness of the network output. The use of the sigmoid function helps the model maintain the output range while determining the proportion of information transfer, while the tanh function helps control and regulate the regularization of the data.

[0100] S33: Real-time analysis and prediction: Input data collected in real time The trained LSTM model is input. The model predicts the propagation performance of electromagnetic waves based on current and past data trends, including electromagnetic wave attenuation, signal attenuation, and interference areas. The output of the LSTM model is interpreted to determine the predicted value of the electromagnetic wave propagation performance and is used to dynamically adjust the communication strategy.

[0101] The spatiotemporal feature maps are included in the input data In the process of combining spatial and temporal characteristics, the rock conductivity and gas concentration are combined with their corresponding spatial position and time information to form a complex input feature vector to capture the temporal and spatial dependencies, including:

[0102] Spatial location encoding: Assign a unique geographic tag to each monitoring point in the mine. For example, based on the layout of the mine, each monitoring point can be labeled according to its specific location in the tunnel (such as the entrance, near the ventilation shaft, mining face, etc.). The geographic tag is converted into a numerical form using one-hot encoding and used as an input feature in the model. For example, the tunnel can be divided into different sections according to its location in the mine, and a vector code can be assigned to each section.

[0103] Time factor integration: The timestamp of data collection is converted into a format that the model can understand. This is achieved by calculating the time interval from the start of monitoring, converting the time into a continuous value that reflects the duration from the start of monitoring to the present. The converted time value is combined with conductivity and gas concentration to form a time series data set for training the LSTM model.

[0104] Combine spatiotemporal features for model training: Combine spatial position coding and time factors with sensor data to form a comprehensive feature vector. For the data at each time point, the feature vector can be [spatial coding, time value, conductivity, gas concentration]. Use the constructed comprehensive feature vector as the input of the LSTM model, and the model simultaneously learns the spatiotemporal dependency in the process of electromagnetic wave propagation.

[0105] The output of the LSTM model is designed to predict relevant indicators of electromagnetic wave propagation, including signal strength, attenuation rate, or the quality of the communication link. By learning patterns and trends in time series data, the model can dynamically predict the electromagnetic wave propagation performance under specific time and space conditions and guide the communication system to make corresponding adjustments.

[0106] S4 specifically includes:

[0107] Receiving LSTM model output: The central processing unit regularly receives prediction data generated by the improved LSTM model, including predictions of electromagnetic wave propagation conditions, signal strength, signal-to-noise ratio, data transmission rate, and potential interruption risks for each monitoring point;

[0108] Calculate the optimal path: Use the dynamic routing algorithm and the prediction results of the LSTM model to calculate the optimal communication path that avoids interference and attenuation areas;

[0109] Dynamically adjust the transmission path: According to the calculated optimal path, dynamically update the routing table and adjust the data transmission path to ensure that the electromagnetic waves are sent through the best path;

[0110] Deploy and adjust wireless relay nodes: According to the dynamically adjusted transmission path, guide the deployment and position adjustment of wireless relay nodes. The wireless relay nodes are equipped with automatic positioning and movement functions. According to the instructions of the central processing unit, adjust the physical position and configuration to support the new optimal communication link.

[0111] The dynamic routing algorithm uses the graph-based Dijkstra algorithm, which includes:

[0112] Mine Propagation Network Modeling: Modeling electromagnetic wave propagation in mines as a weighted graph , where V is the node set in the graph, representing the wireless relay node, and E is the edge set, representing the communication link. There is a weight , representing the node To Node The communication cost or quality of the signal. The weight reflects the interference level and attenuation factor of the signal. The calculation steps of the Dijkstra algorithm are as follows:

[0113] Initialization: For each node in the graph Assign a distance value , for the starting point ; For all other nodes, , create an empty set S to store the nodes whose shortest paths have been found;

[0114] Update distance: for nodes For each adjacent node n of , if n is not in S, update for ;

[0115] Determine the shortest path: Select the node with the smallest distance from the set of unprocessed nodes , add to set S, update node The distance value of the adjacent nodes;

[0116] Repeat updating the distance and determining the shortest path until all nodes are in S, that is, the shortest path from the starting point to all other nodes is found.

[0117] In the present invention, the weight of the edge It not only reflects the traditional distance cost, but also combines the influence of rock conductivity and gas concentration on electromagnetic wave propagation. Therefore, we can define the weight as follows:

[0118] , where PathLoss Represents a slave node arrive Signal path loss, Interference Indicates the interference level on the link, GasLevel Indicates the influence of gas concentration on the link. It is the weight coefficient for adjusting the influence of each factor.

[0119] This weight calculation method enables the Dijkstra algorithm to find the optimal communication path taking into account signal quality, interference and gas concentration. In this way, the mine audio communication system can dynamically adjust its communication path to avoid areas with poor signal quality, high interference or high gas concentration, thereby ensuring the reliability and efficiency of communication.

[0120] Suppose in a mine audio radio wave communication network, there are four communication nodes: A, B, C and D. These nodes are connected through different communication links. The weight of each link reflects the communication quality of the link (taking into account factors such as signal attenuation and interference). It is necessary to calculate the optimal communication path from node A to node D.

[0121] The link weights are set as follows:

[0122] The link weight from A to B is 4 (low interference and attenuation);

[0123] The link weight from A to C is 2 (lowest interference and attenuation);

[0124] The link weight from B to C is 5 (moderate interference and attenuation);

[0125] The link weight from B to D is 10 (higher interference and attenuation);

[0126] The link weight from C to D is 3 (low interference and attenuation);

[0127] Initialization: Set A as the source node, so: , initially, the set S is empty.

[0128] Update distance value: Starting from A:

[0129] Visit node A, consider its adjacent nodes B and C, and update their distance values

[0130] ;

[0131] ; Add A to set S.

[0132] Select the unvisited node with the shortest distance;

[0133] Node C has the shortest distance value, so C is selected as the next node.

[0134] Update the distance value, starting from C:

[0135] Visit node C, consider its adjacent nodes B and D, and update their distance values:

[0136] (The distance value of B remains unchanged);

[0137] , add C to the set S.

[0138] Repeat the update process: Nodes B and D are not visited, the distance value of node B is 4, and that of node D is 5. Select B as the next node.

[0139] To visit node B, we only need to consider node D and update the distance value of D:

[0140] (The distance value of D remains unchanged), add B to the set S.

[0141] Result: All nodes have been processed and the shortest path has been determined. The optimal path from A to D is , the total weight or path cost is 5.

[0142] Through this process, we can see how the Dijkstra algorithm selects the shortest path step by step. Even if some direct paths (such as A to B) seem to have lower costs in the initial stage, the algorithm will still consider the overall shortest path. In this case, even though the path through C to D has a lower total cost, it becomes the final selected path. Such a path selection mechanism can effectively avoid areas with severe interference and attenuation, ensuring the optimization of communication links within the mine.

[0143] The optimization model in S5 uses a Bayesian network model to deal with uncertainty and the probabilistic relationship between variables, selects electromagnetic wave coding strategies in a changing mine environment, constructs a Bayesian network model to represent the impact of rock conductivity and gas concentration on communication performance, and infers the optimal coding strategy.

[0144] Building a Bayesian network model specifically includes:

[0145] Define nodes: Create nodes to represent key variables that affect communication performance, including formation conductivity EDR, gas concentration GC, coding redundancy CR, and error correction level ECL; define the probabilistic dependency between key variables, that is, coding redundancy and error correction level depend on formation conductivity and gas concentration;

[0146] Parameter learning: Collect historical data, including formation conductivity, gas concentration, and the corresponding optimal coding redundancy and error correction level, and use the collected data to estimate the conditional probability distribution of each node in the Bayesian network, for example, estimate the probability of a specific coding redundancy and error correction level under given formation conductivity and gas concentration;

[0147] Reasoning and decision-making: Real-time monitoring of the current rock conductivity and gas concentration. Given the current rock conductivity and gas concentration, Bayesian network is used for reasoning to calculate the posterior probability of different coding scheme parameters (CR and ECL);

[0148] Output recommendation plan: The posterior probability is calculated by Bayes' rule, and the formula is as follows:

[0149] ,in, is the posterior probability of the specific coding redundancy CR and error correction level ECL when the rock conductivity EDR and gas concentration GC are given, is the probability of observing a specific formation conductivity and gas concentration given the encoded parameters, learned from historical data, is the prior probability of coding redundancy and error correction level, is the marginal probability of the currently observed rock conductivity and gas concentration;

[0150] Select the best solution: Select the coding redundancy and error correction level that maximizes the posterior probability as the recommended solution.

[0151] like Figure 2 As shown, a mine audio communication control system is used to implement the above-mentioned mine audio communication control method, including the following modules:

[0152] Rock conductivity monitoring module: It uses a distributed sensor network, deployed in the mine area, to continuously monitor changes in rock conductivity and transmit data to the central processing unit in real time;

[0153] Gas concentration monitoring module: Use gas monitoring equipment, installed in the mine operation area, to detect gas concentration in real time and send distribution data to the central processing unit;

[0154] Data analysis and prediction module: includes a central processing unit equipped with machine learning algorithms to analyze rock conductivity and gas concentration data, predict electromagnetic wave propagation conditions and communication link quality and stability;

[0155] Dynamic routing adjustment module: It includes dynamic routing algorithms and mobile wireless relay nodes. It dynamically adjusts the transmission path according to electromagnetic wave propagation conditions and communication link quality evaluation, and automatically adjusts the location and configuration of wireless relay nodes.

[0156] Coding adjustment module: Based on environmental adaptive signal coding technology, the coding scheme is dynamically adjusted according to the real-time monitoring data in the communication link, and the redundancy and error correction capability of the electromagnetic wave are optimized to adapt to the transmission conditions.

[0157] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0158] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A mine audio communication control method, characterized in that: The following steps are involved: S1: Implement real-time monitoring of rock conductivity, use a sensor network distributed in key areas of coal mines, continuously monitor changes in rock conductivity, and transmit rock conductivity data to a central processing unit in real time; S2: Equip coal seam gas monitoring equipment to detect the gas concentration in each operating area in real time and transmit the gas concentration distribution data to the central processing unit; S3: In the central processing unit, based on the received rock conductivity and gas concentration distribution data, the machine learning algorithm is used to analyze the electromagnetic wave propagation conditions of the current underground environment and predict the quality and stability of the communication link; S4: Implement dynamic routing protocols, dynamically adjust the transmission path according to electromagnetic wave propagation conditions and communication link quality assessment, avoid areas with severe interference and attenuation, and deploy mobile wireless relay nodes. The wireless relay nodes automatically adjust their physical positions and configurations according to the dynamically adjusted transmission paths to build the optimal communication links. S5: Implement environmental adaptive signal coding technology, dynamically adjust the coding scheme according to the real-time monitoring data of rock conductivity and gas concentration distribution in the optimal communication link, and use the optimization model to automatically optimize the redundancy and error correction capability of electromagnetic waves to adapt to the current transmission conditions. The machine learning algorithm in S3 is based on an improved long short-term memory network LSTM model. The central processing unit first pre-processes the received rock conductivity and gas concentration data, uses the LSTM model to analyze the impact of rock conductivity and gas concentration on electromagnetic wave propagation, processes time series data, identifies patterns and trends in the data, and predicts the propagation performance of electromagnetic waves under different underground environmental conditions; Simulate and analyze communication links to predict link quality and stability under current mine conditions, including signal strength, signal-to-noise ratio, data transmission rate, and interruptions. The S4 specifically includes: receiving LSTM model output: the central processing unit regularly receives prediction data generated by the improved LSTM model, including predictions of electromagnetic wave propagation conditions for each monitoring point; Calculate the optimal path: Use the dynamic routing algorithm and the prediction results of the LSTM model to calculate the optimal communication path that avoids interference and attenuation areas; Dynamically adjust the transmission path: According to the calculated optimal path, dynamically update the routing table and adjust the data transmission path to ensure that the electromagnetic waves are sent through the best path; Deploy and adjust wireless relay nodes: According to the dynamically adjusted transmission path, guide the deployment and position adjustment of wireless relay nodes. Wireless relay nodes are equipped with automatic positioning and movement functions. According to the instructions of the central processing unit, adjust the physical position and configuration to support the new optimal communication link. The dynamic routing algorithm uses the graph-based Dijkstra algorithm, which specifically includes: Mine propagation network modeling: modeling the electromagnetic wave propagation in the mine as a weighted graph , where V is the node set in the graph, representing the wireless relay node, and E is the edge set, representing the communication link. There is a weight , representing the node To Node The communication cost or quality of the signal. The weight reflects the interference level and attenuation factor of the signal. The calculation steps of the Dijkstra algorithm are as follows: Initialization: For each node in the graph Assign a distance value , for the starting point ; For all other nodes, , create an empty set S to store the nodes whose shortest paths have been found; Update distance: for nodes For each adjacent node n, if n is not in S, update for ; Determine the shortest path: Select the node with the smallest distance from the set of unprocessed nodes , add to set S, update node The distance value of the adjacent nodes; Repeatedly update the distance and determine the shortest path until all nodes are S The shortest path from the starting point to all other nodes is found.

2. The mine audio communication control method according to claim 1, characterized in that: The S1 includes deploying multi-point conductivity sensors in the main passages and working surfaces of the coal mine to directly measure the conductivity of the rock, and the monitored conductivity data is transmitted to the central processing unit in real time via a wireless network.

3. The mine audio communication control method according to claim 1, characterized in that: The use of the LSTM model to analyze the impact of rock conductivity and gas concentration on electromagnetic wave propagation specifically includes: S31: Feature Engineering: Integrate the multi-dimensional features of rock conductivity, gas concentration distribution, and depth to construct a comprehensive feature set to capture the complex factors affecting the propagation of electromagnetic waves, perform spatiotemporal feature mapping, and integrate spatial position and time factors into the comprehensive feature set to simulate the actual propagation process of electromagnetic waves in mines; S32: Model establishment and training: Build an LSTM model and design an LSTM network architecture, including multiple LSTM units, and the LSTM unit includes an input layer, multiple LSTM layers, and an output layer. Use historical data to train the LSTM model. During training, learn how to predict the propagation performance of electromagnetic waves based on the historical and current values ​​of rock conductivity and gas concentration. S33: Real-time analysis and prediction: Input data collected in real time The trained LSTM model is input. The model predicts the propagation performance of electromagnetic waves based on current and past data trends, including electromagnetic wave attenuation, signal attenuation, and interference areas. The output of the LSTM model is interpreted to determine the predicted value of the electromagnetic wave propagation performance and is used to dynamically adjust the communication strategy.

4. The mine audio communication control method according to claim 3, characterized in that: The spatiotemporal feature map is included in the input data In the process of combining spatial and temporal characteristics, the rock conductivity and gas concentration are combined with their corresponding spatial position and time information to form a complex input feature vector to capture the temporal and spatial dependencies, including: Spatial location encoding: Assign a unique geographic tag to each monitoring point in the mine, use one-hot encoding to convert the geographic tag into a numerical form, and use it as an input feature in the model; Time factor integration: The timestamp of data collection is converted into a format that the model can understand. This is achieved by calculating the time interval from the start of monitoring, converting the time into a continuous value that reflects the duration from the start of monitoring to the present. The converted time value is combined with conductivity and gas concentration to form a time series data set for training the LSTM model. Combine spatiotemporal features for model training: Combine spatial position encoding and time factors with sensor data to form a comprehensive feature vector. Use the constructed comprehensive feature vector as the input of the LSTM model. The model also learns the spatiotemporal dependencies in the process of electromagnetic wave propagation.

5. The mine audio communication control method according to claim 1, characterized in that: The optimization model in S5 adopts a Bayesian network model to process uncertainty and the probabilistic relationship between variables, selects electromagnetic wave coding strategies in a changing mine environment, constructs a Bayesian network model to represent the impact of rock conductivity and gas concentration on communication performance, and infers the optimal coding strategy.

6. The mine audio communication control method according to claim 5, characterized in that: The construction of the Bayesian network model specifically includes: Define nodes: Create nodes to represent key variables that affect communication performance, including formation conductivity EDR, gas concentration GC, coding redundancy CR, and error correction level ECL; define the probabilistic dependency between key variables, that is, coding redundancy and error correction level depend on formation conductivity and gas concentration; Parameter learning: Collect historical data, including formation conductivity, gas concentration, and the corresponding optimal coding redundancy and error correction level, and use the collected data to estimate the conditional probability distribution of each node in the Bayesian network; Reasoning and decision-making: Real-time monitoring of the current rock conductivity and gas concentration. Given the current rock conductivity and gas concentration, Bayesian network is used for reasoning to calculate the posterior probability of different coding scheme parameters. Output recommendation plan: The posterior probability is calculated by Bayes' rule, and the formula is as follows: ,in, is the posterior probability of the specific coding redundancy CR and error correction level ECL when the rock conductivity EDR and gas concentration GC are given, is the probability of observing a specific formation conductivity and gas concentration given the encoded parameters, learned from historical data, is the prior probability of coding redundancy and error correction level, is the marginal probability of the currently observed rock conductivity and gas concentration; Select the best solution: Select the coding redundancy and error correction level that maximizes the posterior probability as the recommended solution.

7. A mine audio communication control system, used to implement the mine audio communication control method according to any one of claims 1 to 6, characterized in that: Includes the following modules: Rock conductivity monitoring module: It uses a distributed sensor network, deployed in the mine area, to continuously monitor changes in rock conductivity and transmit data to the central processing unit in real time; Gas concentration monitoring module: Use gas monitoring equipment, installed in the mine operation area, to detect gas concentration in real time and send distribution data to the central processing unit; Data analysis and prediction module: includes a central processing unit equipped with machine learning algorithms to analyze rock conductivity and gas concentration data, predict electromagnetic wave propagation conditions and communication link quality and stability; Dynamic routing adjustment module: It includes dynamic routing algorithms and mobile wireless relay nodes. It dynamically adjusts the transmission path according to electromagnetic wave propagation conditions and communication link quality evaluation, and automatically adjusts the location and configuration of wireless relay nodes. Coding adjustment module: Based on environmental adaptive signal coding technology, the coding scheme is dynamically adjusted according to the real-time monitoring data in the communication link, and the redundancy and error correction capability of the electromagnetic wave are optimized to adapt to the transmission conditions.