Communication method and system of tunnel construction heterogeneous equipment based on federal learning

Abstract

The invention discloses a federated learning-based communication method and system for tunnel construction heterogeneous equipment, and the method comprises the steps: creating and training a channel distribution model on all equipment participating in tunnel construction, and transmitting model parameters to a central server; the central server updates a global model based on channel allocation model parameters of the devices participating in tunnel construction and allocates the global model to the devices participating in tunnel construction; and each piece of equipment participating in tunnel construction evaluates a current channel by adopting a global model, and selects an optimal channel as a data transmission channel. According to the method, the federated learning technology is utilized, distributed model training among equipment is achieved, data privacy and safety are ensured, channel conflicts and interference can be effectively reduced through dynamic channel allocation, and the data transmission rate is increased.

Description

Technical Field

[0001] The present invention relates to the field of tunnel construction communication technology, and in particular to a communication method and system for heterogeneous tunnel construction equipment based on federated learning. Background Art

[0002] In modern tunnel construction, with the expansion of project scale and the complexity of construction environment, the widespread application of heterogeneous equipment has become a trend. These heterogeneous equipment include various machinery, sensors, monitoring instruments and automatic control systems, which work together to improve construction efficiency and safety. However, these devices face multiple challenges in the process of information transmission and communication.

[0003] In tunnel construction, the current communication technology solutions mainly include static channel allocation and fixed encryption protocols. These technical solutions meet the basic communication needs to a certain extent, but they have the following significant disadvantages in complex construction environments:

[0004] Inefficient communication: Existing static channel allocation methods usually fix the channel allocation strategy at the design stage, and fail to consider the dynamic changes in the number and status of equipment during the construction process. This results in channel resources being over-allocated or under-allocated in actual operations, resulting in frequent signal conflicts and interference. Since the equipment cannot adjust channel usage according to real-time conditions, communication delays increase and the overall data transmission rate decreases, far below actual requirements. In complex tunnel environments, signals are affected by terrain, equipment layout, and other interference factors, resulting in unstable communications. Existing solutions fail to effectively deal with these interferences, resulting in increased data packet loss rates, affecting the real-time and accuracy of construction information.

[0005] Lack of flexibility: There are many types of equipment used in construction, and each device may use different communication protocols. Existing technical solutions usually lack unified standards, leading to obstacles in information sharing and data exchange, resulting in low efficiency in collaborative operations between devices. Current technical solutions cannot dynamically adjust communication strategies based on equipment status, environmental changes, or construction progress, resulting in insufficient communication efficiency in different construction stages and conditions, limiting the project's flexibility.

[0006] Insufficient security: Existing encryption methods usually use fixed algorithms and levels, and cannot adjust encryption strategies based on the sensitivity of communication data or real-time risks. This lack of flexibility may lead to security vulnerabilities in data transmission, especially in multi-device, multi-user environments. Data is prone to privacy leakage during transmission, especially when centralized transmission is used. Since multiple devices need to share data, this process exposes sensitive information in the network, increasing the possibility of attack. Summary of the invention

[0007] The purpose of the present invention is to provide a communication method and system for heterogeneous equipment in tunnel construction based on federated learning, which solves the problems of low communication efficiency, insufficient security and poor flexibility in the prior art through dynamic channel allocation and multi-level data encryption mechanism, thereby improving the communication performance and security of heterogeneous equipment in tunnel construction.

[0008] The present invention provides a communication method for heterogeneous tunnel construction equipment based on federated learning, comprising:

[0009] Create and train a channel allocation model on all devices involved in tunnel construction, and transmit the trained model parameters to a central server;

[0010] The central server updates the global model based on the trained model parameters and distributes them to each device involved in the tunnel construction;

[0011] Each device involved in tunnel construction uses a global model to evaluate the current channel and select the best channel as the data transmission channel.

[0012] Preferably, the creating and training of a channel allocation model on all devices involved in tunnel construction includes:

[0013] Each device locally collects historical data related to channel usage, including signal strength, transmission rate, channel status, and environmental parameters;

[0014] Considering that the equipment in the construction environment belongs to different brands and models, horizontal federated learning is used to create a channel allocation model;

[0015] A machine learning algorithm is selected to train the created channel allocation model until convergence, to obtain a trained channel allocation model and save the model parameters.

[0016] Preferably, the method comprises:

[0017] The created channel allocation model is trained using a linear regression algorithm, and the model parameters are updated using a gradient descent method.

[0018] Preferably, the training further includes:

[0019] Organize and clean the collected data, as well as classify the data and label different devices, environments, and time periods.

[0020] Preferably, updating the global model means,

[0021] The channel allocation model parameters of each device involved in tunnel construction are weighted averaged to form a global model.

[0022] Preferably, each device involved in the tunnel construction uses a global model to evaluate the current channel and selects the best channel as the data transmission channel, including:

[0023] Each device involved in tunnel construction monitors the status of the current channel in real time, including signal strength, interference level and transmission rate;

[0024] The device uses the updated global model to evaluate the status of the current channel and predict the availability of the current channel;

[0025] According to the prediction results, the best channel is selected as the data transmission channel.

[0026] Preferably, selecting the best channel as the data transmission channel according to the prediction result includes:

[0027] Select the channel with the highest frequency of occurrence as the data transmission channel;

[0028] If the frequencies are tied, the signal with the smaller number will be selected first.

[0029] Preferably, the trained model parameters are transmitted to the central server in an encrypted manner, including:

[0030] Each device selects encryption algorithms of different encryption levels according to the type of data to be transmitted, and encrypts the data to be transmitted; the data type includes sensitive data and ordinary data; the sensitive data is data related to cost or operation content, including: model parameters, equipment location, operation time, operation route; the ordinary data is data other than sensitive data;

[0031] The encrypted data is transmitted through the allocated channel.

[0032] Preferably, the method further comprises:

[0033] During the construction process, the equipment continuously monitors the channel usage and data transmission status, and regularly uploads feedback information to the central server; the feedback content should include the channel usage efficiency and data transmission stability;

[0034] The central server optimizes the channel allocation model and encryption strategy according to the information fed back by the device; the optimization includes readjusting the parameters of the channel allocation model and adjusting the encryption level of the encryption process.

[0035] The present invention also provides a communication system for heterogeneous tunnel construction equipment based on federated learning, which is used to implement the communication method for heterogeneous tunnel construction equipment based on federated learning. The system includes: equipment involved in tunnel construction, a central server and a communication module.

[0036] Configuration on the equipment involved in tunnel construction:

[0037] A data collection module, used to locally collect historical and real-time data related to channel usage;

[0038] Model training module, used to create and train a channel allocation model using horizontal federated learning;

[0039] The channel allocation module is used to monitor the status of the current channel in real time, including signal strength, interference level and transmission rate; use the updated global model to evaluate the status of the current channel and predict the availability of the current channel; based on the prediction results, select the best channel as the data transmission channel;

[0040] The central server is used to receive local channel allocation model parameters of each device involved in tunnel construction, store and update the global model and issue it;

[0041] The communication module is used to realize data interaction between the equipment involved in tunnel construction and the central server.

[0042] Preferably, the equipment involved in tunnel construction is also equipped with:

[0043] The encryption module is used to select encryption algorithms of different encryption levels according to the type of data to be transmitted, encrypt the data to be transmitted, and transmit the encrypted data through the allocated channel.

[0044] Preferably, the model training module has the following training process:

[0045] A linear regression algorithm is used to calculate the predicted channel allocation results;

[0046] The loss function is calculated based on the prediction results, and the model parameters are updated using the gradient descent method. Then the next round of training is carried out until convergence to obtain the trained model parameters.

[0047] Preferably, the central server is also configured with a firewall and security protocol.

[0048] Preferably, each device involved in tunnel construction is equipped with a wireless communication module.

[0049] The beneficial effects of the present invention are:

[0050] The present invention uses federated learning technology to achieve distributed model training between devices, ensuring data privacy and security, while improving the adaptability of the model through local training. At the same time, the present invention uses feedback information to optimize the model and strategy to achieve self-adjustment and continuous improvement. The present invention realizes efficient communication and data sharing of multiple heterogeneous devices, ensuring that different types of equipment can work together seamlessly and improving overall construction efficiency.

[0051] The present invention adopts real-time channel evaluation and selection algorithm to dynamically allocate channels according to environmental changes and device status, thereby improving communication efficiency and stability.

[0052] During the data transmission process, the present invention dynamically selects the encryption level according to the data characteristics and the transmission environment, thereby ensuring the security of the data during the transmission process and reducing the risk of information leakage. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A schematic flow chart of a communication method for heterogeneous tunnel construction equipment based on federated learning provided in an embodiment of the present invention;

[0054] Figure 2 A schematic diagram of a channel allocation model training based on federated learning provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments and the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0056] It should also be noted that, in order to avoid obscuring the present invention due to unnecessary details, only structures and / or processing steps closely related to the solutions according to the present invention are shown in the accompanying drawings, while other details that are not closely related to the present invention are omitted.

[0057] It should be emphasized that the term “include / comprises” when used herein refers to the presence of features, elements, steps or components, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0058] It should also be noted that, unless otherwise specified, the term “connection” herein may refer not only to a direct connection but also to an indirect connection with an intermediate.

[0059] It should be emphasized here that the step marks mentioned below are not intended to limit the order of the steps, but it should be understood that the steps can be executed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be executed simultaneously.

[0060] The first embodiment of the present invention provides a communication method for heterogeneous tunnel construction equipment based on federated learning, see Figure 1 , including equipment preparation, system configuration, initial data collection, model training and updating, dynamic channel allocation, data encryption and transmission, as well as feedback and optimization. The detailed description is as follows:

[0061] S1. Local device configuration, including:

[0062] S11. Tunnel construction equipment configuration:

[0063] Ensure that all equipment involved in tunnel construction (such as excavators, sensors, monitoring systems, etc.) have the necessary communication functions to achieve real-time data collection and transmission.

[0064] Each device needs to be equipped with appropriate computing resources to support local training of the federated learning model. It is recommended to use hardware with sufficient processing power and memory, such as industrial-grade computers or smart terminals.

[0065] S12. Communication module configuration:

[0066] All equipment involved in tunnel construction should be equipped with wireless communication modules (such as LTE, 5G, Wi-Fi, etc.) to ensure efficient and stable data communication between devices.

[0067] Make sure the device supports multiple communication protocols so that it can quickly switch channels and protocols when dynamic channel allocation is used.

[0068] S13. Software installation and debugging:

[0069] Build an edge cloud computing center on local devices and install necessary programs on them, including data collection programs, a local model training framework based on federated learning, and an encryption module for data transmission.

[0070] Configure training parameters such as learning rate, number of training epochs, and local training batch size to optimize model performance.

[0071] Perform system debugging to ensure that all functions are functioning properly and the device can be stably connected to the central server.

[0072] S2. Central server configuration

[0073] S21. Central server construction:

[0074] A central server is established to receive local model parameters of each device involved in tunnel construction, store and update the global model, and coordinate communication between devices.

[0075] It should be noted that the local model refers to the channel allocation model that adapts to changes in the construction environment and is built based on local historical data related to channel usage by each device and based on federated learning technology;

[0076] The global model refers to a model formed by weighted averaging of the local model parameters of each device involved in tunnel construction.

[0077] The central server needs to have high processing power and storage capacity to support multiple devices to process data and update models simultaneously.

[0078] S22. Safety configuration:

[0079] Firewalls and security protocols are used to ensure the security of data transmission between local devices and central servers.

[0080] Ensure that all transmitted data is encrypted to prevent data leakage or tampering.

[0081] S3. Initial data collection

[0082] S31. Data collection plan development:

[0083] Develop an initial data collection plan to clearly define the data collection objectives, data types (such as signal strength, environmental parameters, device status, etc.), and data sampling frequency.

[0084] Determine the data collection period so that sufficient data can be collected in the early stages of construction.

[0085] S32, Equipment Data Collection:

[0086] Each device collects relevant data in real time according to the collection plan, recording information such as channel usage, transmission rate, and environmental factors.

[0087] During the data collection process, ensure that the connection between the device and the central server is unobstructed so that the collected data can be uploaded in real time.

[0088] S33. Data storage and organization:

[0089] The collected data is stored in the edge cloud computing center of the local device and sorted and cleaned to ensure data quality.

[0090] Classify data and label different devices, environments, and time periods to facilitate subsequent model training.

[0091] S4. Local channel allocation model training and global model update

[0092] S41. Local channel allocation model training:

[0093] Each device uses the historical data related to channel usage collected locally to train based on the selected machine learning algorithm and saves the model parameters after each round of training. The training process includes data preprocessing, model initialization and training process.

[0094] S42. Upload local channel allocation model parameters:

[0095] After the local channel allocation model training is completed, the edge cloud computing center uploads the updated model parameters (instead of the original data) to the central server.

[0096] It should be noted that the model parameter uploading process should use encrypted transmission to ensure that the data is not leaked or tampered with during transmission.

[0097] S43, global model update:

[0098] After the central server receives the model parameters uploaded by all devices, it updates the global model.

[0099] The specific method is to perform weighted averaging on the model parameters of each device to form a new global model.

[0100] The updated global model will be sent to all devices to ensure that the devices use the latest model in subsequent channel allocation.

[0101] S5. Dynamic channel allocation based on global model

[0102] S51, channel assessment:

[0103] During use, each device evaluates the status of the current channel in real time, including parameters such as signal strength, interference level and transmission rate.

[0104] The device uses the updated global model to evaluate the status of the current channel and predict the availability and interference level of the current channel.

[0105] S52, Best channel selection:

[0106] Based on the evaluation results, the device selects the best channel for data transmission. The selection process should take into account the communication requirements and environmental conditions of the device. It should be noted that encryption mechanism is used for encryption before data transmission.

[0107] S53, channel allocation feedback:

[0108] The central server feeds back the selected channels and their usage to each device to form a closed-loop feedback mechanism to ensure continuous optimization of models and strategies.

[0109] See also Figure 2 In this embodiment, the training process of the local channel allocation model is as follows:

[0110] The core of dynamic channel allocation is to use federated learning technology to build a channel allocation model that adapts to changes in the construction environment. The details are as follows:

[0111] S411. Data collection:

[0112] Each device locally collects historical data related to channel usage, including signal strength, transmission rate, channel status, and environmental factors (such as humidity, temperature, etc.).

[0113] Considering that the construction machinery in the construction environment belongs to different brands and models, the characteristics of their perception data are quite different. Each data set shares the same sample ID, environmental features, and equipment feature space, but the samples are different. Therefore, horizontal federated learning is used to build training models and evaluation models. An honest third-party collaborator C is introduced, usually an edge cloud center server. A and B are construction machinery from different brands in tunnel construction. Aggregate all environmental features and equipment features in the working environment into a complete feature set, based on privacy protection, calculate training losses and gradients, and make full use of data from both parties to train the model.

[0114] ,

[0115] in, is the feature space (including device features and environment features), is the label space (i.e., feature name), is the sample ID space (i.e., the construction machinery number), and the three constitute the complete training data set .matrix Indicates each data owner Holds data whose rows represent the same sample and whose columns represent the same feature.

[0116] S412, local model training:

[0117] Multiple devices with the same data structure collaborate to learn the machine learning model through the edge cloud computing center. The model uses a linear regression algorithm. Linear regression function:

[0118] ,

[0119] in, is the independent variable, representing input features, such as device status, environmental parameters, etc. and are model parameters, The prediction results for channel assignment.

[0120] The loss function is defined as the mean squared error between the predicted value and the true value:

[0121] ,

[0122] in, The actual result of channel allocation is is the number of data samples.

[0123] The core optimization goal is:

[0124] ,

[0125] Derived separately to get the gradient function:

[0126] ,

[0127] ,

[0128] Let the derivative be zero, and the optimal solution is:

[0129] ,

[0130] ,

[0131] S413, model parameter update: After each round of training, the device calculates the training gradient and uses the gradient descent method to update the model parameters:

[0132] ,

[0133] ,

[0134] The currently selected gradient is masked using encryption technology and the encrypted result is sent to the central server. Each device transmits model parameters instead of raw data to protect privacy.

[0135] S414: If the channel allocation model predicts that information source, then:

[0136] ,

[0137] is the regression coefficient, and the prediction result is a column vector consisting of all sample regression results.

[0138] In the embodiment of the present invention, the best channel is selected in the following manner:

[0139] The best channel is selected according to the evaluation results of the global model. If the best channel is generated as a column vector by multiple factors, the channel with the highest frequency of occurrence is selected as the transmission channel. If the frequencies of occurrence are the same, the one with the smaller sequence number is selected first.

[0140] In the embodiment of the present invention, during the uploading process of the local channel allocation model parameters, encrypted transmission is adopted, and the encryption process is as follows:

[0141] S421, Encryption Strategy Construction

[0142] This embodiment uses a variety of encryption algorithms based on data characteristics and communication environment, as follows:

[0143] Data characteristic identification: The device analyzes the characteristics of the data to be transmitted before data transmission, including the data type (sensitive data, ordinary data) and the current communication transmission environment (including security risks and bandwidth limitations).

[0144] In this embodiment, sensitive data includes model parameters, equipment location, operation time, operation route, etc., which are highly related to customer costs or related to operation content. Ordinary data includes engine trouble-free working time, vehicle status data (such as turn signal, power arm lifting signal), etc., which are less related to cost and operation content.

[0145] Encryption level selection: Based on the data characteristics, the device selects the appropriate encryption algorithm and level to ensure high communication efficiency while ensuring data security:

[0146] For sensitive data, use a higher encryption level (such as AES-256).

[0147] For normal data, a lower encryption level (such as AES-128) is used.

[0148] S422, the dynamic encryption process is as follows,

[0149] Before data transmission, the data is encrypted. The formula is as follows:

[0150] ,

[0151] in, For the encrypted data, is the encryption algorithm, is the original data, Is the encryption key.

[0152] The encrypted data is transmitted through the allocated channel, at which point the device will report the integrity and availability of the encrypted information back to the central server for monitoring.

[0153] After receiving the data, the receiver decrypts it according to the corresponding decryption key. The formula is as follows:

[0154] ,

[0155] in, is the decryption algorithm.

[0156] S423, Dynamic Adjustment of Encryption Strategy

[0157] The device can adjust encryption strategies in real time based on real-time environmental changes and communication risks. If it detects security risks in the communication environment, the device can automatically increase the encryption level to ensure data security.

[0158] During the implementation of the above technical solution, the channel allocation model and encryption strategy are continuously fed back and optimized as follows:

[0159] A. During the construction process, the equipment continuously monitors the channel usage and data transmission status, and regularly uploads feedback information to the central server.

[0160] The feedback should include the efficiency of channel utilization, the stability of data transmission and possible problems.

[0161] B. The central server optimizes the channel allocation model and encryption strategy based on the information fed back by the equipment to ensure that the system can continue to operate efficiently in a constantly changing construction environment.

[0162] The optimization process should include the readjustment of model parameters and the dynamic update of encryption algorithms to meet the needs of different construction stages.

[0163] Another embodiment of the present invention further provides a tunnel construction heterogeneous equipment communication system for implementing the above technical solution, comprising: equipment involved in tunnel construction, a central server and a communication module,

[0164] In this embodiment, each device involved in tunnel construction needs to be equipped with appropriate computing resources to support the training of the local channel allocation model. It is recommended to use hardware with sufficient processing power and memory, such as an industrial-grade computer or smart terminal.

[0165] Each device is also configured with:

[0166] Data collection module, used to collect historical data related to channel usage locally, including signal strength, transmission rate, channel status and environmental factors (such as humidity, temperature, etc.);

[0167] The model training module is used to construct a channel allocation training model and an evaluation model using horizontal federated learning. The training process is as follows: a linear regression algorithm is used to calculate the predicted channel allocation results; a loss function is calculated based on the predicted results, and a gradient function is derived and calculated. The gradient descent method is used to update the model parameters, and then the next round of training is carried out until convergence; after the training is completed, the transmission model parameters are sent to the central server through the encryption module;

[0168] The channel allocation module is used to evaluate the status of the current channel in real time, including parameters such as signal strength, interference level and transmission rate; use the updated global model to evaluate the current channel and predict the availability and stability of the channel; and select the best channel for data transmission based on the evaluation results;

[0169] The encryption module is used to select different encryption algorithms to encrypt the data to be transmitted according to the data type.

[0170] It should be noted that the encryption mechanism of the encryption module is the same as that of the above method embodiment, and will not be described again.

[0171] It should be noted that it is necessary to ensure that each device supports multiple communication protocols so that channels and protocols can be quickly switched during dynamic channel allocation.

[0172] In this embodiment, the central server is used to receive local channel allocation model parameters of each device involved in tunnel construction, store and update the global model, and coordinate communication between devices.

[0173] Specifically, after receiving the model parameters uploaded by all devices, the central server updates the global model by weighted averaging the model parameters of each device to form a new global model. The updated global model will be sent to all devices to ensure that the devices use the latest model in subsequent channel allocation.

[0174] It should be noted that the central server needs to have high processing power and storage capacity to support multiple devices to perform data processing and model updates simultaneously.

[0175] It should be noted that the central server is also configured with firewalls and security protocols to ensure the security of data transmission between the device and the central server.

[0176] In this embodiment, each device involved in the tunnel construction is equipped with a wireless communication module (such as LTE, 5G, Wi-Fi, etc.) to realize data interaction between the device involved in the tunnel construction and the central server.

[0177] It should be further explained that the system of this embodiment is regularly maintained and updated to ensure that the software versions of all devices and central servers are kept up to date, thereby improving the security and stability of the system.

[0178] When necessary, conduct a comprehensive evaluation of the system to ensure that it adapts to the development of new technologies and changes in the construction environment.

[0179] This embodiment adopts the above-mentioned technical means, and can effectively reduce channel conflicts and interference through dynamic channel allocation, improve data transmission rate, and multi-level data encryption mechanism ensures the security of data during transmission, reducing the risk of data leakage. By dynamically selecting the encryption level, it is possible to maintain high communication efficiency while ensuring security. The present invention can flexibly adjust channels and encryption strategies according to real-time environmental changes, adapt to a variety of construction scenarios, and improve the adaptability and stability of the system.

[0180] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A communication method for heterogeneous equipment in tunnel construction based on federated learning, characterized in that: include: Create and train a channel allocation model on all devices involved in tunnel construction, and transmit the trained model parameters to a central server; The central server updates the global model based on the trained model parameters and distributes them to each device involved in the tunnel construction; Each device involved in tunnel construction uses a global model to evaluate the current channel and select the best channel as the data transmission channel.

2. According to claim 1, a communication method for heterogeneous tunnel construction equipment based on federated learning is characterized in that: The channel allocation model is created and trained on all devices involved in tunnel construction, including: Each device locally collects historical data related to channel usage, including signal strength, transmission rate, channel status, and environmental parameters; Considering that the equipment in the construction environment belongs to different brands and models, horizontal federated learning is used to create a channel allocation model; A machine learning algorithm is selected to train the created channel allocation model until convergence, to obtain a trained channel allocation model and save the model parameters.

3. According to claim 2, a communication method for heterogeneous tunnel construction equipment based on federated learning is characterized in that: The method comprises: The created channel allocation model is trained using a linear regression algorithm, and the model parameters are updated using a gradient descent method.

4. According to claim 3, a communication method for heterogeneous tunnel construction equipment based on federated learning is characterized in that: The training also includes: Organize and clean the collected data, as well as classify the data and label different devices, environments, and time periods.

5. According to claim 2, a communication method for heterogeneous tunnel construction equipment based on federated learning is characterized in that: The updating of the global model means, The channel allocation model parameters of each device involved in tunnel construction are weighted averaged to form a global model.

6. A method for communicating heterogeneous equipment in tunnel construction based on federated learning according to claim 5, characterized in that: Each device involved in the tunnel construction uses a global model to evaluate the current channel and selects the best channel as the data transmission channel, including: Each device involved in tunnel construction monitors the status of the current channel in real time, including signal strength, interference level and transmission rate; The device uses the updated global model to evaluate the status of the current channel and predict the availability of the current channel; According to the prediction results, the best channel is selected as the data transmission channel.

7. A method for communicating heterogeneous equipment in tunnel construction based on federated learning according to claim 6, characterized in that: The step of selecting the best channel as the data transmission channel according to the prediction result includes: Select the channel with the highest frequency of occurrence as the data transmission channel; If the frequencies are tied, the signal with the smaller number will be selected first.

8. The communication method for heterogeneous tunnel construction equipment based on federated learning according to claim 1, characterized in that: The trained model parameters are transmitted to the central server in an encrypted manner, including: Each device selects encryption algorithms of different encryption levels according to the type of data to be transmitted, and encrypts the data to be transmitted; the data type includes sensitive data and ordinary data; the sensitive data is data related to cost or operation content, including: model parameters, equipment location, operation time, operation route; the ordinary data is data other than sensitive data; The encrypted data is transmitted through the allocated channel.

9. A method for communicating heterogeneous equipment in tunnel construction based on federated learning according to claim 8, characterized in that: The method further comprises: During the construction process, the equipment continuously monitors the channel usage and data transmission status, and regularly uploads feedback information to the central server; the feedback content should include the channel usage efficiency and data transmission stability; The central server optimizes the channel allocation model and encryption strategy according to the information fed back by the device; the optimization includes readjusting the parameters of the channel allocation model and adjusting the encryption level of the encryption process.

10. A communication system for heterogeneous equipment in tunnel construction based on federated learning, characterized in that: A method for implementing a communication method for heterogeneous tunnel construction equipment based on federated learning as described in any one of claims 1 to 9, the system comprising: equipment involved in tunnel construction, a central server and a communication module, Configuration on the equipment involved in tunnel construction: A data collection module, used to locally collect historical and real-time data related to channel usage; Model training module, used to create and train a channel allocation model using horizontal federated learning; The channel allocation module is used to monitor the status of the current channel in real time, including signal strength, interference level and transmission rate; use the updated global model to evaluate the status of the current channel and predict the availability of the current channel; based on the prediction results, select the best channel as the data transmission channel; The central server is used to receive local channel allocation model parameters of each device involved in tunnel construction, store and update the global model and issue it; The communication module is used to realize data interaction between the equipment involved in tunnel construction and the central server.

11. A communication system for heterogeneous tunnel construction equipment based on federated learning according to claim 10, characterized in that: The equipment involved in the tunnel construction is also equipped with: The encryption module is used to select encryption algorithms of different encryption levels according to the type of data to be transmitted, encrypt the data to be transmitted, and transmit the encrypted data through the allocated channel.

12. A communication system for heterogeneous tunnel construction equipment based on federated learning according to claim 10, characterized in that: The model training module, the training process is as follows: A linear regression algorithm is used to calculate the predicted channel allocation results; The loss function is calculated based on the prediction results, and the model parameters are updated using the gradient descent method. Then the next round of training is carried out until convergence to obtain the trained model parameters.

13. A communication system for heterogeneous tunnel construction equipment based on federated learning according to claim 10, characterized in that: The central server is also configured with firewalls and security protocols.

14. A communication system for heterogeneous tunnel construction equipment based on federated learning according to claim 10, characterized in that: Each device involved in tunnel construction is equipped with a wireless communication module.

Images