A railway tunnel wireless coverage system and method based on relay technology

Abstract

The present invention relates to the technical field of railway wireless communication systems, and discloses a railway tunnel wireless coverage system and method based on relay technology, wherein a railway tunnel wireless coverage method based on relay technology includes the following steps: obtaining comprehensive train operation information and historical train communication demand information; constructing the comprehensive train operation information into ordered structure data, and constructing the historical train communication demand information into time series data; inputting the structure data and time series data into a wireless optimization model, and outputting a railway tunnel wireless coverage optimization strategy. The present invention can improve the signal strength and stability in the railway tunnel, and ensure smooth wireless communication of the train during tunnel operation.

Description

Technical Field

[0001] The present invention relates to the technical field of railway wireless communication systems, and more specifically, to a railway tunnel wireless coverage system and method based on relay technology. Background Art

[0002] With the rapid development of railway transportation, trains are running more and more frequently in tunnels, and the demand for wireless communication is also increasing. In tunnels, since signal propagation is affected by various factors, such as the structure, length, terrain changes and speed of the tunnel, the quality of wireless communication between the train and the ground is often difficult to guarantee. This problem not only affects the communication experience of passengers inside the train, but also affects the safety monitoring and dispatching management of train operation. Therefore, how to achieve effective wireless coverage in railway tunnels and ensure smooth communication of trains in tunnels has become an important issue that the industry needs to solve urgently.

[0003] At present, the existing wireless coverage solutions for railway tunnels mainly rely on the deployment of fixed base stations and signal amplification equipment, but these methods are difficult to adapt to the complex changes in tunnel environment and the needs of constantly moving trains. At the same time, traditional wireless coverage optimization methods are mostly single-mode information processing, lacking comprehensive consideration of train operation status, tunnel environment and historical communication needs. This one-sided optimization method not only makes it difficult to effectively improve the quality of signal coverage, but also limits the rational use of resources. Therefore, to improve the wireless coverage effect in railway tunnels, it is necessary to combine the real-time operation data of trains and historical communication needs, and realize intelligent wireless resource management and scheduling through data analysis and model optimization. In this context, a railway tunnel wireless coverage method based on relay technology came into being. By comprehensively considering various factors of train operation, a wireless optimization model based on real-time data input is established, which effectively solves the challenges of wireless communication in tunnels and meets the communication needs of trains under dynamic changes in tunnel operation. Summary of the invention

[0004] The present invention provides a railway tunnel wireless coverage system and method based on relay technology, which solves the technical problems in related technologies such as weak wireless signals in tunnels, unstable communications, and failure to meet the real-time communication needs of trains in tunnels.

[0005] The present invention provides a railway tunnel wireless coverage method based on relay technology, comprising the following steps:

[0006] Step 100, obtaining comprehensive train operation information, which includes: a comprehensive train operation description text composed of tunnel environment information, train operation information, and wireless network information; obtaining historical train communication demand information;

[0007] Step 200, constructing the comprehensive train operation information into ordered structured data, the structured data includes multiple data segments, each data segment includes name data and volume data, one name data corresponds to one object, one name data is linked to one volume data, the volume data linked to the name data is the feature representation of the object referred to by the name data when the data segment is collected, and the objects include: tunnel, train, wireless network;

[0008] Step 300, constructing time series data based on historical train communication demand information, where the time series data includes multiple time series units;

[0009] Step 400, input the structural data and the time series data into the wireless optimization model, the wireless optimization model includes a first backbone network, the first backbone network includes a first data structure recognition layer, a second data structure recognition layer, a third data structure recognition layer, a fourth data structure recognition layer, a first data structure recognition fusion layer, a first fully connected layer, a second fully connected layer, and a third fully connected layer; wherein the structural data is input into the first data structure recognition layer, and the first hidden feature is output to the second data structure recognition layer; the second hidden feature output by the second data structure recognition layer is output to the first data structure recognition fusion layer; the first data structure recognition fusion layer outputs the first fused hidden feature to the first fully connected layer; the first fully connected layer outputs the railway tunnel wireless coverage optimization strategy.

[0010] In a preferred embodiment, one name data is linked to one index list, the value of the element in the index list represents the relationship between the object referred to by the name data and all objects existing in reality, and one element in the index list is linked to one name data.

[0011] In a preferred embodiment, the characteristic representation of the tunnel includes: tunnel length, tunnel cross section, tunnel wall material, tunnel curvature, tunnel slope;

[0012] The train characteristics include: train position, train speed, train length, and train communication requirements. The train communication requirements include the data rate, number of connections, and service type required for communication.

[0013] The characteristics of the wireless network include: wireless base station information and historical channel data.

[0014] In a preferred embodiment, the relationships existing in reality include:

[0015] The train runs in the tunnel, and the train's position and speed are related to the length and curvature of the tunnel.

[0016] In a preferred embodiment, the method for constructing time series data includes:

[0017] The historical communication demand information of the train is extracted, and the features of the extracted historical communication demand information of the train are divided into a plurality of time sequence units according to a time window, and the order of the time sequence units is consistent with the time order of the collection period of the historical communication demand information of the train.

[0018] In a preferred embodiment, the train's historical communication demand information includes: a train's historical communication demand description text consisting of historical communication traffic data at different locations in the tunnel, historical communication service quality data at different locations in the tunnel, historical communication service type data at different locations in the tunnel, and historical communication demand data at different time periods in the tunnel.

[0019] In a preferred embodiment, the comprehensive train operation information and the historical train communication demand information are input into the wireless optimization model after feature engineering;

[0020] The comprehensive train operation information includes a comprehensive train operation description text, and the train historical communication demand information includes a train historical communication demand description text; the train operation comprehensive description text and the train historical communication demand description text are used as a feature engineering method through the WordEmbedding algorithm.

[0021] In a preferred embodiment, the railway tunnel wireless coverage optimization strategy includes the optimized location selection of base station deployment in the tunnel, the transmission power control of the base station in the tunnel, the direction adjustment of the base station antenna in the tunnel, and the quantitative evaluation index of the base station coverage quality in the tunnel.

[0022] In a preferred embodiment, a reinforcement learning method is used to train the first backbone network in the wireless optimization model;

[0023] The second fully connected layer is pre-trained separately to train the wireless optimization model to have the ability to accurately coordinate the interference between base stations in the tunnel, ensuring that the use of network communication when the train is traveling in the tunnel is not affected by the interference between base stations in the tunnel;

[0024] The third fully connected layer is pre-trained separately to train the wireless optimization model to have the ability to accurately predict the communication needs of trains while traveling in tunnels.

[0025] In a preferred embodiment, a railway tunnel wireless coverage system based on relay technology includes the following modules:

[0026] Data acquisition module: used for comprehensive train operation information and train historical communication demand information;

[0027] Data structuring module: constructs the comprehensive information of train operation into ordered structured data, and constructs time series data based on the historical communication demand information of the train;

[0028] The data processing module inputs the structural data and time series data into the wireless optimization model and outputs the wireless coverage optimization strategy for the railway tunnel;

[0029] The wireless optimization module ensures the normal communication needs of passengers on trains traveling in tunnels based on the wireless coverage optimization strategy of railway tunnels.

[0030] The beneficial effects of the present invention are:

[0031] Enhanced wireless signal coverage: The wireless coverage system based on relay technology can effectively improve the signal strength and stability in railway tunnels, ensuring unimpeded wireless communication during train operation in tunnels. This improvement enhances the communication experience of passengers.

[0032] Intelligent resource optimization: Combining the real-time train operation data and historical communication needs, through the application of data structuring and optimization models, the present invention can intelligently configure and manage wireless resources to achieve precise base station deployment and transmission power control. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flow chart of a railway tunnel wireless coverage method based on relay technology of the present invention;

[0034] Figure 2 is an example of a set of data samples of the present invention;

[0035] Figure 3 is an example of another set of data samples of the present invention. DETAILED DESCRIPTION

[0036] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the present specification. Various examples may omit, replace, or add various processes or components as needed. In addition, the features described in some examples may also be combined in other examples.

[0037] At least one embodiment of the present invention discloses a railway tunnel wireless coverage method based on relay technology, such as Figure 1 As shown, the following steps are included:

[0038] Step 100, obtaining comprehensive train operation information, which includes: a comprehensive train operation description text composed of tunnel environment information, train operation information, and wireless network information; obtaining historical train communication demand information;

[0039] In one embodiment of the present invention, the comprehensive train operation information is collected every 3 minutes when the train starts, and the collection of the comprehensive train operation information is stopped when the train arrives at the terminal.

[0040] Step 200, constructing the comprehensive train operation information into ordered structured data, the structured data includes multiple data segments, each data segment includes name data and volume data, one name data corresponds to one object, one name data is linked to one volume data, the volume data linked to the name data is the feature representation of the object referred to by the name data when the data segment is collected, and the objects include: tunnel, train, wireless network;

[0041] In one embodiment of the present invention, one name data is linked to one index list, the value of the element in the index list indicates the relationship between the object referred to by the name data and all objects existing in reality, and one element in the index list is linked to one name data;

[0042] The elements of the index list contain 0 and non-0 elements. If the i-th element is a non-0 element, it means that the name data corresponding to the index list has a relationship with the i-th name data in reality.

[0043] In one embodiment of the present invention, the characteristic representation of the tunnel includes: tunnel length, tunnel cross section, tunnel wall material, tunnel curvature, tunnel slope;

[0044] The train characteristics include: train location, train speed, train length, and train communication requirements. Train communication requirements include the data rate required for communication, the number of connections, and the type of service (such as voice, video, data, etc.);

[0045] The characteristic representation of the wireless network includes: wireless base station information, historical channel data;

[0046] Wireless base station information includes: the deployment location of the wireless base station in the tunnel, transmission power, operating frequency, antenna type used (such as omnidirectional antenna, directional antenna, etc.), antenna height, base station protocol (such as GSM, UMTS, LTE, etc.), frequency planning (such as frequency planning and coordination of public networks, private networks, broadcasting, etc.);

[0047] Historical channel data include: channel fading (such as large-scale fading and small-scale fading), signal-to-noise ratio (such as received signal strength, noise power, etc.), interference level (such as co-channel interference, adjacent channel interference, etc.), delay spread (such as maximum lead delay, maximum lag delay, root mean square delay spread, etc.).

[0048] In one embodiment of the present invention, the relationships existing in reality include:

[0049] The train runs in the tunnel, and the train's position and speed are related to the length and curvature of the tunnel.

[0050] Step 300, constructing time series data based on historical train communication demand information, where the time series data includes multiple time series units;

[0051] A structured approach includes:

[0052] The historical communication demand information of the train is extracted, and the features of the extracted historical communication demand information of the train are divided into a plurality of time sequence units according to a time window, and the order of the time sequence units is consistent with the time order of the collection period of the historical communication demand information of the train.

[0053] In one embodiment of the present invention, the collection period of the historical train communication demand information is the same as the collection period of obtaining the comprehensive train operation information.

[0054] In one embodiment of the present invention, the historical communication demand information of the train includes: historical communication traffic data of the train at different positions in the tunnel (such as uplink and downlink data rates), historical communication service quality data of the train at different positions in the tunnel (such as delay, packet loss rate, connection establishment success rate, etc.), historical communication service type data of the train at different positions in the tunnel (such as the proportion of voice calls, video streaming, Web browsing, file downloads, etc.), and historical communication demand data of the train at different time periods in the tunnel (such as the difference in communication traffic and service quality between peak hours and off-peak hours), which constitute a train historical communication demand description text.

[0055] In one embodiment of the present invention, comprehensive train operation information and historical train communication demand information are input into the wireless optimization model after feature engineering;

[0056] The comprehensive train operation information includes a comprehensive train operation description text, and the train historical communication demand information includes a train historical communication demand description text; the train operation comprehensive description text and the train historical communication demand description text are used as a feature engineering method through the WordEmbedding algorithm.

[0057] Step 400, inputting the structure data and the time series data into the wireless optimization model, the wireless optimization model includes a first backbone network, the first backbone network includes a first data structure identification layer, a second data structure identification layer, a third data structure identification layer, a fourth data structure identification layer, a first data structure identification fusion layer, a first fully connected layer, a second fully connected layer, and a third fully connected layer; wherein the structure data is input into the first data structure identification layer, and the first hidden feature is output to the second data structure identification layer; the second hidden feature output by the second data structure identification layer is output to the first data structure identification fusion layer; the first data structure identification fusion layer outputs the first fused hidden feature to the first fully connected layer; the first fully connected layer outputs the railway tunnel wireless coverage optimization strategy;

[0058] The structure data is input into the third data structure recognition layer, and the third hidden feature is output to the first data structure recognition fusion layer and the second fully connected layer, and the second fully connected layer outputs the interference coordination result between the base stations in the tunnel;

[0059] The time series data is input into the fourth data structure recognition layer, and the fourth hidden feature is output to the first data structure recognition fusion layer and the third fully connected layer. The third fully connected layer outputs the communication demand prediction of the train in the tunnel.

[0060] In one embodiment of the present invention, the calculation formula of the first data structure identification layer is as follows:

[0061]

[0062] in, The first data structure identifies the first layer. The structure data of the layer object hidden features of objects, Indicates Layer object hidden features of objects, and Respectively represent and object and objects A collection of objects with object relationships. Indicates the first The weight matrix of the layer, represents the normalization function, represents the sigmoid activation function, , Indicates the total number of layers, hour, , represents the wth individual data, hour, Representation Object The first hidden feature of .

[0063] In one embodiment of the present invention, the calculation formula of the second data structure identification layer is as follows:

[0064]

[0065] in, and represents the activation vector of the reset gate and update gate at step t; represents the candidate state generated in step t; represents the second hidden feature of the t-th step, represents the second hidden feature of the t-1th step; A, B, D, E, F, G represent the transformation matrices of the first, second, third, fourth, fifth, and sixth trainable parameters respectively; c, d, g represent the first, second, and third trainable bias parameters; , represents the first hidden feature of the object v output when the t-th data segment is input into the first data structure recognition layer, , The number of data fragments representing the structure data, hour, , Represents the collection of all objects; represents the hyperbolic tangent activation function; Represents the sigmoid activation function.

[0066] In one embodiment of the present invention, the calculation formula of the third data structure identification layer is as follows:

[0067]

[0068] in, Indicates the first Layer object hidden features of objects, Indicates Layer object hidden features of objects, and Respectively represent and object and objects A collection of objects with object relationships. Indicates the first The weight matrix of the layer, represents the normalization function, represents the sigmoid activation function, , Indicates the total number of layers, hour, , Indicates Individual data, hour, Representation Object The third hidden feature of .

[0069] In one embodiment of the present invention, the calculation formula of the fourth data structure identification layer is as follows:

[0070]

[0071] in, and Indicates The activation vectors of the reset gate and update gate of the step; Indicates The candidate states generated by the step; Indicates The fourth hidden feature of the step, Indicates The fourth hidden feature of the step; , , , , , , respectively represent the transformation matrices of the seventh, eighth, ninth, tenth, eleventh, and twelfth trainable parameters; , , represents the fourth, fifth, and sixth trainable bias parameters; Indicates the first Timing units, , The number of time series units representing the time series data; represents the hyperbolic tangent activation function; Represents the sigmoid activation function.

[0072] In one embodiment of the present invention, the calculation formula for the first data structure to identify the fusion layer is as follows:

[0073]

[0074] in, represents the first fusion hidden feature, represents the feature combination function (concatenation function or summation function), Indicates The second hidden feature of the step, Representation Object The third hidden feature of Represents the collection of all objects, Indicates A fourth hidden feature.

[0075] In one embodiment of the present invention, the calculation formula of the first fully connected layer is as follows:

[0076]

[0077] in, represents the first output vector, and its fth component value represents the probability value of the fth railway tunnel wireless coverage optimization strategy. The railway tunnel wireless coverage optimization strategy with the largest probability value is selected as the output. The railway tunnel wireless coverage optimization strategy group contains all executable railway tunnel wireless coverage optimization strategies. represents the first fusion hidden feature, Indicates full connection.

[0078] In one embodiment of the present invention, the railway tunnel wireless coverage optimization strategy includes the optimized location selection of base station deployment in the tunnel, the transmission power control of the base station in the tunnel, the direction adjustment of the base station antenna in the tunnel, and the quantitative evaluation index of the coverage quality of the base station in the tunnel.

[0079] In one embodiment of the present invention, the calculation formula of the second fully connected layer is as follows:

[0080]

[0081] in, represents the second output vector, whose The component represents the The interference coordination result of each base station. For example, a result of 0 indicates that the base station does not adopt interference coordination, a result of -1 indicates that the base station adopts power control, and a result of 1 indicates that the base station adopts frequency planning, etc. Representation Object The third hidden feature of Indicates full connection.

[0082] In one embodiment of the present invention, the calculation formula of the third fully connected layer is as follows:

[0083]

[0084] in, Represents the first output matrix, the first row of the matrix The column indicates the number of tunnels the train passes through. The data rate requirement for the time period is The column indicates the number of tunnels the train passes through. The number of network connections in the time period, the third row of the matrix The column indicates the number of tunnels the train passes through. The business proportion in each time period indicates the rate demand proportion of voice, video and data services. Indicates The fourth hidden feature, Indicates full connection.

[0085] In one embodiment of the present invention, a reinforcement learning method is used to train the first backbone network in the wireless optimization model, such as Q learning.

[0086] In one embodiment of the present invention, the second fully connected layer is pre-trained separately, and the training wireless optimization model has the ability to accurately coordinate the interference between base stations in the tunnel, ensuring that the use of network communication when the train is traveling in the tunnel is not affected by the interference between the base stations in the tunnel.

[0087] In one embodiment of the present invention, the third fully connected layer is pre-trained separately, and the training wireless optimization model has the ability to accurately predict the communication needs of the train during the tunnel driving process. The training loss function calculation formula is as follows:

[0088]

[0089] in represents the communication demand prediction loss value, represents the total number of time periods that the train travels in the tunnel, represents the actual data rate requirement in the qth time period, represents the data rate requirement for the qth time period output by the third fully connected layer, Indicates the actual number of connections required in the qth time period, Indicates the number of connections required for the qth time period of the output of the third fully connected layer.

[0090] In one embodiment of the present invention, a specific example of the above-mentioned railway tunnel wireless coverage method based on relay technology is provided:

[0091] This example focuses on a busy high-speed railway line in China, especially a long-distance railway tunnel passing through a mountainous area (limited to a specific section). The train flow in this section is large, including high-speed passenger trains and freight trains, so the quality of wireless communication is required to be high. The main cities in the study area include Hangzhou, Quzhou, etc., which have close transportation links and are important economic and logistics hubs.

[0092] Data preparation:

[0093] like Figure 2 As shown, comprehensive train operation information:

[0094] Tunnel environment information: tunnel length, cross section, wall material, curvature, slope.

[0095] Train operation information: train location, speed, length, communication requirements (data rate, number of connections, service type).

[0096] Wireless network information: base station deployment location, transmission power, operating frequency, antenna type, height, network architecture, protocol, frequency planning; historical channel data such as channel fading, signal-to-noise ratio, interference level, delay spread, etc.

[0097] like Figure 3 As shown, the train's historical communication demand information:

[0098] Historical communication traffic data (uplink and downlink data rates), historical communication service quality data (latency, packet loss rate, connection establishment success rate), historical communication service type data (the proportion of voice calls, video streaming, web browsing, file downloads, etc.), historical communication demand data in different time periods (the difference between peak hours and off-peak hours).

[0099] Data annotation:

[0100] Based on the train's historical communication event data and expert knowledge, the data is annotated to mark the time periods and areas with poor communication quality. For example, when a train loses signal or experiences severe delays at a specific tunnel location, the location and time point are marked as areas that need to be optimized;

[0101] Model training (pre-training)

[0102] Through pre-training, the model has the ability to coordinate interference between base stations in the tunnel and accurately predict the communication needs of trains while traveling in the tunnel, thereby optimizing the wireless coverage model;

[0103] Model training (reinforcement learning)

[0104] Reinforcement Learning Framework:

[0105] Environment: Actual railway tunnel environment, taking into account factors such as complex tunnel structure and high-speed train movement.

[0106] Intelligent agent: wireless optimization system, responsible for adjusting the wireless coverage optimization strategy of railway tunnels.

[0107] Status: current comprehensive train operation information, train historical communication demand information, etc.

[0108] Action: Railway tunnel wireless coverage optimization strategy.

[0109] Rewards: Rewards are given based on the improved signal coverage quality and user satisfaction, such as successfully improving signal strength and reducing call drop rate.

[0110] The reward function is as follows:

[0111]

[0112] in, Indicates reward, Indicates the signal coverage quality after taking action. Indicates the signal coverage quality before taking action; Expected quality of service, The actual service quality achieved, which is expressed as the maximum allowed delay and the minimum allowed packet loss rate; Represents the total available bandwidth resources, Indicates the actual bandwidth resources used.

[0113] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation mode. The above-mentioned specific implementation mode is merely illustrative and not restrictive. Under the guidance of this embodiment, ordinary technicians in this field can also make more forms of equivalent embodiments, all of which are within the protection of this embodiment.

Claims

1. A railway tunnel wireless coverage method based on relay technology, characterized in that: The following steps are involved: Step 100, obtaining comprehensive train operation information, which includes: a comprehensive train operation description text composed of tunnel environment information, train operation information, and wireless network information; obtaining historical train communication demand information; Step 200, constructing the comprehensive train operation information into ordered structured data, the structured data includes multiple data segments, each data segment includes name data and volume data, one name data corresponds to one object, one name data is linked to one volume data, the volume data linked to the name data is the feature representation of the object referred to by the name data when the data segment is collected, and the objects include: tunnel, train, wireless network; Step 300, constructing time series data based on historical train communication demand information, where the time series data includes multiple time series units; Step 400, input the structural data and the time series data into the wireless optimization model, the wireless optimization model includes a first backbone network, the first backbone network includes a first data structure recognition layer, a second data structure recognition layer, a third data structure recognition layer, a fourth data structure recognition layer, a first data structure recognition fusion layer, a first fully connected layer, a second fully connected layer, and a third fully connected layer; wherein the structural data is input into the first data structure recognition layer, and the first hidden feature is output to the second data structure recognition layer; the second hidden feature output by the second data structure recognition layer is output to the first data structure recognition fusion layer; the first data structure recognition fusion layer outputs the first fused hidden feature to the first fully connected layer; the first fully connected layer outputs the railway tunnel wireless coverage optimization strategy.

2. According to claim 1, a railway tunnel wireless coverage method based on relay technology is characterized in that: One name data is linked to one index list. The value of the element in the index list indicates the relationship between the object referred to by the name data and all objects existing in reality. One element in the index list is linked to one name data.

3. According to claim 2, a railway tunnel wireless coverage method based on relay technology is characterized in that: The characteristics of the tunnel include: tunnel length, tunnel cross section, tunnel wall material, tunnel curvature, tunnel slope; The train characteristics include: train position, train speed, train length, and train communication requirements. The train communication requirements include the data rate, number of connections, and service type required for communication. The characteristics of the wireless network include: wireless base station information and historical channel data.

4. According to claim 2, a railway tunnel wireless coverage method based on relay technology is characterized in that: The relationships that exist in reality include: The train runs in the tunnel, and the train's position and speed are related to the length and curvature of the tunnel.

5. The method for wireless coverage of railway tunnels based on relay technology according to claim 1, characterized in that: Methods for constructing time series data include: The historical communication demand information of the train is extracted, and the features of the extracted historical communication demand information of the train are divided into a plurality of time sequence units according to a time window, and the order of the time sequence units is consistent with the time order of the collection period of the historical communication demand information of the train.

6. The method for wireless coverage of railway tunnels based on relay technology according to claim 1, characterized in that: The train's historical communication demand information includes: a train's historical communication demand description text consisting of historical communication flow data at different positions of the train in the tunnel, historical communication service quality data at different positions of the train in the tunnel, historical communication service type data at different positions of the train in the tunnel, and historical communication demand data at different time periods in the tunnel.

7. The method for wireless coverage of railway tunnels based on relay technology according to claim 1, characterized in that: The comprehensive train operation information and the historical train communication demand information are input into the wireless optimization model through feature engineering; The comprehensive train operation information includes a comprehensive train operation description text, and the train historical communication demand information includes a train historical communication demand description text; the train operation comprehensive description text and the train historical communication demand description text are used as a feature engineering method through the WordEmbedding algorithm.

8. The method for wireless coverage of railway tunnels based on relay technology according to claim 1, characterized in that: The railway tunnel wireless coverage optimization strategy includes the optimized location selection of base station deployment in the tunnel, the transmission power control of the base station in the tunnel, the direction adjustment of the base station antenna in the tunnel, and the quantitative evaluation index of the base station coverage quality in the tunnel.

9. The method for wireless coverage of railway tunnels based on relay technology according to claim 1, characterized in that: The first backbone network in the wireless optimization model is trained by using a reinforcement learning method; The second fully connected layer is pre-trained separately to train the wireless optimization model to have the ability to accurately coordinate the interference between base stations in the tunnel, ensuring that the use of network communication when the train is traveling in the tunnel is not affected by the interference between base stations in the tunnel; The third fully connected layer is pre-trained separately to train the wireless optimization model to have the ability to accurately predict the communication needs of trains while traveling in tunnels.

10. A railway tunnel wireless coverage system based on relay technology, characterized in that: Includes the following modules: Data acquisition module: used to obtain comprehensive train operation information, including: comprehensive train operation description text composed of tunnel environment information, train operation information, and wireless network information; obtain historical train communication demand information; Data structuring module: used to construct the comprehensive information of train operation into ordered structured data. The structured data includes multiple data segments. Each data segment includes name data and volume data. One name data corresponds to one object. One name data is linked to one volume data. The volume data linked to the name data is the feature representation of the object referred to by the name data when the data segment is collected. The objects include: tunnels, trains, and wireless networks. The historical communication demand information of the train is constructed into time series data. The time series data contains multiple time series units. The data processing module is used to input the structural data and the time series data into the wireless optimization model, wherein the wireless optimization model includes a first backbone network, and the first backbone network includes a first data structure recognition layer, a second data structure recognition layer, a third data structure recognition layer, a fourth data structure recognition layer, a first data structure recognition fusion layer, a first fully connected layer, a second fully connected layer, and a third fully connected layer; wherein the structural data is input into the first data structure recognition layer, and the first hidden feature is output to the second data structure recognition layer; the second hidden feature output by the second data structure recognition layer is output to the first data structure recognition fusion layer; the first data structure recognition fusion layer outputs the first fused hidden feature to the first fully connected layer; the first fully connected layer outputs the railway tunnel wireless coverage optimization strategy; The wireless optimization module ensures the normal communication needs of passengers on trains traveling in tunnels based on the wireless coverage optimization strategy of railway tunnels.

Images