Seismic wave acquisition system based on SDN and fusion routing algorithm

Abstract

The invention discloses a seismic wave acquisition system based on an SDN and a fusion routing algorithm, the seismic wave acquisition system comprises an SDA controller, a fusion routing intelligent closed-loop control framework, a DQL-HRA routing algorithm module and a software architecture, the fusion routing intelligent closed-loop control framework is deployed in the SDA controller, the fusion routing algorithm module is arranged in the fusion routing intelligent closed-loop control framework, and the DQL-HRA routing algorithm module is arranged in the software architecture. The fusion routing intelligent closed-loop control framework comprises a unified state sensing and preprocessing module, a cooperative control intelligent decision module and a control strategy generation and deployment module. The SDN controller and the fusion routing algorithm are combined, the network state change can be dynamically adapted, the purpose of dynamically adjusting and optimizing the routing path is achieved, and through dynamic path optimization, data packet loss and retransmission are reduced, node energy consumption is balanced, local overload is avoided, so that the data transmission efficiency is improved, and the use requirement is met.

Description

Technical Field

[0001] The present invention relates to the technical field of seismic wave acquisition, and particularly to a seismic wave acquisition system based on SDN and a fusion routing algorithm. Background Art

[0002] Seismic wave detection has important applications in underground space monitoring. However, it faces challenges such as complex geological environments, multi-path signal attenuation, dynamic topology changes, and limited node energy. Traditional wireless sensor network (WSN) routing algorithms (such as centralized or distributed) have insufficient global optimization capabilities and local optimum problems, resulting in low data transmission efficiency and unbalanced energy consumption, making it unable to meet the demand for efficient data transmission. Considering the above situation, the present application proposes a seismic wave acquisition system based on SDN and a fusion routing algorithm. Summary of the Invention

[0003] Based on the technical problems existing in the background art, the present invention proposes a seismic wave acquisition system based on SDN and a fusion routing algorithm.

[0004] The seismic wave acquisition system based on SDN and a fusion routing algorithm proposed by the present invention includes an SDA controller, a fusion routing intelligent closed-loop control framework, a DQL-HRA routing algorithm module, and a software architecture. The fusion routing intelligent closed-loop control framework is deployed in the SDA controller. The fusion routing intelligent closed-loop control framework has a built-in fusion routing algorithm module. The fusion routing intelligent closed-loop control framework includes a unified state perception and preprocessing module, a collaborative control intelligent decision-making module, and a control strategy generation and deployment module;

[0005] The DQL-HRA routing algorithm module includes a centralized load balancing reward function and a distributed load balancing reward function;

[0006] The software architecture includes a service layer, a front-end layer, a back-end layer, a database layer, and a distributed deployment layer.

[0007] Preferably, the unified state perception and preprocessing module has functions of global state perception, data preprocessing, and policy feedback update. Through the global view function of SDN, it can collect network global state information in real time, sense and summarize the state information from forwarding nodes. At the same time, the unified state perception and preprocessing module preprocesses the collected data to provide high-quality input data for subsequent modules, improving the accuracy and reliability of decision-making. The unified state perception and preprocessing module analyzes the transmission results of completed tasks in a timely manner, optimizes and updates the next-round routing strategy in a timely manner, and realizes closed-loop control and dynamic adjustment;

[0008] The network global state information includes network node load, link quality, and remaining energy data.

[0009] Preferably, the collaborative control intelligent decision-making module has functions of dynamic optimization of the reward function, collaborative decision-making of global and local paths, and task priority management. Combining reinforcement learning and collaborative control functions, it dynamically adjusts the weights of centralized and distributed reward functions to meet the multi-objective optimization requirements. At the same time, it can achieve global path optimization and local path adjustment. Using the reinforcement learning algorithm, it optimizes the weight parameters according to the real-time network state to ensure that the routing selection takes into account the goals of energy consumption balance, load distribution, and transmission delay. Global path optimization is oriented to high-priority tasks to generate the optimal path to ensure high reliability and low latency of transmission; local path adjustment is aimed at the dynamic changes of the network, quickly providing a path correction plan, dynamically allocating path resources according to the priority of tasks, high-priority tasks preferentially select the global path, and ordinary tasks mainly rely on the local path. Finally, the collaborative control intelligent decision-making module outputs a comprehensive routing strategy of global and local for subsequent modules to execute and deploy.

[0010] Preferably, the control strategy generation and deployment module has functions of routing strategy generation and flow table deployment. According to the comprehensive routing strategy generated by the collaborative control intelligent decision-making module driven by reinforcement learning, it converts the path selection scheme into a specific flow table. The flow table includes the scheduling rules of global and local paths, supports information annotation of task priorities, and ensures that high-priority tasks are preferentially guaranteed. Through the policy deployment interface of SDN, this module quickly distributes the generated flow table to each forwarding node in the network to guide the efficient forwarding and transmission execution of data traffic, ensuring the efficient transmission of different types of tasks and the rational utilization of resources.

[0011] Preferably, the centralized load balancing reward function is used to achieve network energy consumption balance, load balance, minimization of transmission delay, and path selection stability. Its function formula is as follows:

[0012] R global =λ1·Energy_Balance - λ2·Load_Variance - λ3·Delay + λ4·R history ;

[0013] where R global is the global reward function, Energy_Balance is the energy balance, Load_Variance is the load variance, Delay is the path transmission delay, and R history is the historical reward smoothing term;

[0014] Energy Balance reflects the degree of balance in the remaining energy distribution of each node on the path. By measuring the balance of energy consumption of nodes on the path, this metric guides the routing selection to favor nodes with more remaining energy, thereby dispersing the overall network energy consumption, avoiding premature failure of a certain node due to energy exhaustion, and extending the network lifecycle. The formula it uses is:

[0015] where |P| is the number of nodes on the path, is the remaining energy of node i, is the initial energy of node i;

[0016] Load Variance is used to measure the balance of loads on each node in the network. The smaller the load variance, the more evenly distributed the network load is, effectively reducing the data transmission bottleneck caused by an overloaded single node, and at the same time improving the stability and reliability of network transmission. The formula it uses is:

[0017] where L i is the current load of node i, is the average value of the loads of all nodes in the network, and |N| is the total number of nodes in the network;

[0018] Path Transmission Delay reflects the cumulative transmission delay of data packets from the source node to the destination node. Path transmission delay is one of the key metrics for evaluating network performance. Especially for high-priority tasks, a lower path transmission delay can ensure the real-time nature and transmission efficiency of data, helping to meet the requirements of delay-sensitive applications. The formula it uses is: where, d j is the transmission delay of link j on the path;

[0019] The historical reward smoothing term (R history ) is used to smooth the dynamic changes between the current reward and the historical reward. Introducing the historical reward smoothing term can effectively suppress the frequent changes in path selection caused by short-term reward fluctuations, improve the stability of routing decisions, and ensure that the path planning can achieve a smooth transition under long-term optimization goals. The formula it uses is: where k is the time window size of the historical reward, and R t represents the reward value at time t.

[0020] Preferably, the distributed load balancing reward function depends on the local information of the nodes, dynamically evaluates the load status, energy consumption level, and link quality of neighbor nodes in real time, is applicable to data transmission and path adjustment of ordinary tasks, and can quickly adapt to the dynamic changes of the network. The formula it uses is:

[0021] R local = μ1·ICATT i + μ2·Energy i - μ3·Load i - μ4·PLR i + μ5·R history ;

[0022] Wherein, ICATT i is the link load and interference level, Energy i is the energy consumption status of the node, Load i is the load status of the node, PLR i is the link packet loss rate;

[0023] ICATT i measures the load pressure and interference status of the current path by comprehensively considering the load and link interference of neighbor nodes. By optimizing the load and interference of the link, the transmission bottleneck can be effectively reduced and the data transmission quality can be improved. The formula it uses is: Where N i is the set of neighbor nodes of node i, L j is the load of neighbor node j, T j is the transmission rate of the neighbor node, NL j is the length of the send queue of the neighbor node;

[0024] Energy i is used to measure the proportion of the remaining energy of the node in the total energy. This parameter ensures that nodes with high remaining energy are preferentially selected, which helps to evenly distribute the overall energy consumption of the network and extend the network lifetime. The formula it uses is: Where, is the remaining energy of node i, is the initial energy of node i;

[0025] Load i is used to measure the load status of the node. The greater the node load, the lower the reward obtained. By dynamically evaluating the load of the node, the overload phenomenon of high-load nodes can be effectively avoided, thereby improving the stability and balance of the network. The formula it uses is: Where L i is the current load of the node, T i is the actual transmission rate of the node, NL i represents the length of the queue of the node's send interface;

[0026] PLR iUsed to measure the reliability of the current link. The higher the packet loss rate, the greater the penalty for the reward. By reducing the packet loss rate, the reliability and success rate of data transmission can be effectively improved, ensuring the stable operation of network performance. The formula used is:

[0027] Preferably, the service layer includes a command control module, a response module, a status monitoring module, a data display module, and a data storage module. The command control module is responsible for managing and distributing the operation requirements of users; the response module processes data interaction between the device and the system; the status monitoring module realizes real-time detection of the device operation status; the data display module provides data visualization; the data storage module ensures the secure storage of historical data and log information;

[0028] The front-end layer is developed based on Web technology and includes a data display interface, a system settings interface, a device management interface, a real-time acquisition interface, and a log interface. The front-end layer not only displays the results of business logic processing but also can receive user input, complete system configuration and instruction issuance through interface operations. Moreover, the data displayed by the front-end layer is dynamically updated through a unified data interface to ensure the real-time and accuracy of the data;

[0029] The back-end layer is developed based on the Spring Boot framework and includes a device communication module, a log and monitoring module, a message queue module, a configuration management module, a data processing module, and a cache management module. The device communication module is responsible for accessing device data, completing data parsing and status feedback; the configuration management module supports the maintenance of system configuration parameters, network topology, and historical configurations; the data processing module is responsible for parsing and storing the collected seismic data; the log and monitoring module realizes the recording of system error logs, user operation logs, and operation status monitoring; the cache management module uses Redis to provide an efficient data caching service; the message queue module realizes asynchronous message passing through RabbitMQ to improve task processing efficiency.

[0030] Preferably, the database layer mainly uses a MySQL relational database, which is responsible for storing device information, collected data, system configurations, user operation records, and log information. It includes a device information table, a collection data table, a configuration management table, a log record table, and an operation record table to ensure efficient data query and management. The database supports querying and exporting historical data, ensuring the integrity and traceability of data during long-term storage and use. Through reasonable index design and backup strategies, the database ensures the performance and security in high-concurrency access scenarios;

[0031] The distributed deployment layer is responsible for the load balancing of system services, supports data synchronization and status monitoring, and provides reliable operation guarantee for backend services and data storage. It includes Docker containerization and Spring Boot microservice architecture. Through the splitting of Spring Boot microservice architecture, each functional module of the system can be independently deployed and managed, enhancing the scalability and stability of the system.

[0032] Compared with the existing technologies, the beneficial effects of the present invention are as follows:

[0033] By combining the SDN controller with the fusion routing algorithm, the present invention can dynamically adapt to network state changes, achieve the purpose of dynamically adjusting and optimizing the routing path, reduce packet loss and retransmission through dynamic path optimization, balance node energy consumption, avoid local overload, thereby improving data transmission efficiency and meeting usage requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a block diagram of the seismic wave acquisition system based on SDN and fusion routing algorithm proposed by the present invention;

[0035] Figure 2 is a software architecture diagram of the seismic wave acquisition system based on SDN and fusion routing algorithm proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The present invention will be further explained below with reference to specific embodiments.

[0037] Embodiment

[0038] Refer to Figure 1-2 , this embodiment proposes a seismic wave acquisition system based on SDN and fusion routing algorithm, including an SDA controller, a fusion routing intelligent closed-loop control framework, a DQL-HRA routing algorithm module, and a software architecture. The fusion routing intelligent closed-loop control framework is deployed in the SDA controller, and the fusion routing intelligent closed-loop control framework incorporates a fusion routing algorithm module. The fusion routing intelligent closed-loop control framework includes a unified state perception and preprocessing module, a collaborative control intelligent decision-making module, and a control strategy generation and deployment module;

[0039] Among them, the unified state perception and preprocessing module has functions of global state perception, data preprocessing, and policy feedback and update. Through the global view function of SDN, it can collect network global state information in real time, sense and summarize the state information from forwarding nodes. At the same time, the unified state perception and preprocessing module preprocesses the collected data to provide high-quality input data for subsequent modules, improving the accuracy and reliability of decision-making. The unified state perception and preprocessing module analyzes the transmission results of the completed tasks, timely optimizes and updates the next round of routing policies, and realizes closed-loop control and dynamic adjustment;

[0040] The network global state information includes network node load, link quality, and remaining energy data;

[0041] The cooperative control intelligent decision-making module has functions of dynamic optimization of the reward function, collaborative decision-making of global and local paths, and task priority management. Combining reinforcement learning and cooperative control functions, it dynamically adjusts the weights of centralized and distributed reward functions to meet the multi-objective optimization requirements. At the same time, it can achieve global path optimization and local path adjustment. Using the reinforcement learning algorithm, it optimizes the weight parameters according to the real-time network state to ensure that the routing selection takes into account the goals of energy consumption balance, load distribution, and transmission delay. The global path optimization is for high-priority tasks to generate the optimal path to ensure high reliability and low latency of transmission; the local path adjustment is for the dynamic changes of the network, quickly providing a path correction plan, dynamically allocating path resources according to the priority of tasks, high-priority tasks preferentially select the global path, and ordinary tasks mainly rely on the local path. Finally, the cooperative control intelligent decision-making module outputs the comprehensive routing strategy of global and local for subsequent modules to execute and deploy;

[0042] The control strategy generation and deployment module has functions of routing strategy generation and flow table deployment. According to the comprehensive routing strategy generated by the cooperative control intelligent decision-making module driven by reinforcement learning, it converts the path selection scheme into a specific flow table. The flow table includes the scheduling rules of global and local paths, supports the information annotation of task priorities, and ensures that high-priority tasks are preferentially guaranteed. Through the policy deployment interface of SDN, this module quickly distributes the generated flow table to each forwarding node in the network to guide the efficient forwarding and transmission execution of data traffic, ensuring the efficient transmission of different types of tasks and the reasonable utilization of resources;

[0043] The DQL-HRA routing algorithm module includes a centralized load balancing reward function and a distributed load balancing reward function;

[0044] Among them, the centralized load balancing reward function is used to achieve network energy consumption balance, load balance, minimization of transmission delay, and path selection stability. Its function formula is as follows:

[0045] R global = λ1·Energy_Balance - λ2·Load_Variance - λ3·Delay + λ4·R history ;

[0046] Among them, R global is the global reward function, Energy_Balance is the energy balance, Load_Variance is the load variance, Delay is the path transmission delay, and R history is the historical reward smoothing term;

[0047] Energy_Balance reflects the degree of balance in the distribution of the remaining energy of each node on the path. By measuring the balance of energy consumption of nodes on the path, this metric guides the routing selection to favor nodes with more remaining energy, thereby dispersing the overall network energy consumption, avoiding premature failure of a certain node due to energy exhaustion, and extending the network lifetime. The formula it uses is:

[0048] where |P| is the number of nodes on the path, is the remaining energy of node i, is the initial energy of node i;

[0049] Load_Variance is used to measure the balance of loads of each node in the network. The smaller the load variance, the more evenly distributed the network load is, effectively reducing the data transmission bottleneck caused by an overloaded single node, and at the same time improving the stability and reliability of network transmission. The formula it uses is:

[0050] where L i is the current load of node i, is the average value of the loads of all nodes in the network, and |N| is the total number of nodes in the network;

[0051] Delay reflects the cumulative transmission delay of data packets from the source node to the destination node. Path transmission delay is one of the key metrics for evaluating network performance. Especially for high-priority tasks, a lower path transmission delay can ensure the real-time nature and transmission efficiency of data, helping to meet the requirements of delay-sensitive applications. The formula it uses is: where d j is the transmission delay of link j on the path;

[0052] The historical reward smoothing term (R history ) is used to smooth the dynamic changes between the current reward and the historical reward. Introducing the historical reward smoothing term can effectively suppress the frequent changes in path selection caused by short-term reward fluctuations, improve the stability of routing decisions, and ensure that the path planning can achieve a smooth transition under the long-term optimization goal. The formula it uses is: where k is the time window size of the historical reward, and R t represents the reward value at time t;

[0053] The distributed load balancing reward function depends on the local information of nodes, dynamically evaluates the load status, energy consumption level, and link quality of neighbor nodes in real time, is suitable for data transmission and path adjustment of ordinary tasks, and can quickly adapt to the dynamic changes of the network. The formula it uses is:

[0054] R local = μ1·ICATT i + μ2·Energy i - μ3·Load i - μ4·PLR i + μ5·R history ;

[0055] Wherein, ICATT i is the link load and interference level, Energy i is the energy consumption status of the node, Load i is the load status of the node, PLR i is the link packet loss rate;

[0056] ICATT i measures the load pressure and interference status of the current path by comprehensively considering the load of neighbor nodes and link interference. By optimizing the load and interference of the link, the transmission bottleneck can be effectively reduced and the data transmission quality can be improved. The formula it uses is: Where N i is the set of neighbor nodes of node i, L j is the load of neighbor node j, T j is the transmission rate of the neighbor node, NL j is the length of the send queue of the neighbor node;

[0057] Energy i is used to measure the proportion of the remaining energy of the node in the total energy. This parameter ensures that nodes with high remaining energy are preferentially selected, which helps to evenly distribute the overall energy consumption of the network and extend the network life cycle. The formula it uses is: Where, is the remaining energy of node i, is the initial energy of node i;

[0058] Load i is used to measure the load status of the node. The greater the node load, the lower the reward obtained. By dynamically evaluating the load of the node, the overload phenomenon of high-load nodes can be effectively avoided, thereby improving the stability and balance of the network. The formula it uses is: Where L i is the current load of the node, T i is the actual transmission rate of the node, NL i represents the length of the queue of the node's send interface;

[0059] PLR iIt is used to measure the reliability of the current link. The higher the packet loss rate, the greater the penalty for the reward. By reducing the packet loss rate, the reliability and success rate of data transmission can be effectively improved, ensuring the stable operation of network performance. The formula it uses is:

[0060] The software architecture includes a business layer, a front-end layer, a back-end layer, a database layer, and a distributed deployment layer;

[0061] The business layer includes a command control module, a response module, a status monitoring module, a data display module, and a data storage module. Among them, the command control module is responsible for managing and distributing the operation requirements of users; the response module processes the data interaction between the device and the system; the status monitoring module realizes the real-time detection of the device operation status; the data display module provides data visualization display; the data storage module ensures the secure storage of historical data and log information;

[0062] The front-end layer is developed based on Web technology. It includes a data display interface, a system settings interface, a device management interface, a real-time acquisition interface, and a log interface. The front-end layer not only displays the results of business logic processing, but also can receive user input, complete system configuration and instruction issuance through interface operations. Moreover, the data displayed by the front-end layer is dynamically updated through a unified data interface to ensure the real-time and accuracy of the data;

[0063] In addition, the system settings interface is used to complete the settings of system parameters. It provides the function of dynamically configuring core parameters such as the server IP address, port number, acquisition mode, and sampling frequency, and also supports historical configuration management. Users can view or import the system parameter files saved historically. In addition, the function setting module of the system settings interface provides comprehensive management of the data file storage path, performance parameters, and node configuration to ensure the flexibility and configurability of the system operation;

[0064] The device management interface realizes the device management function. It establishes connections with each acquisition node through network communication and displays detailed information such as the IP address, port number, longitude and latitude, device status, and battery power of the device in real time. The system further supports network topology visualization, dynamically displays the connection relationship between devices and the real-time status of network nodes. At the same time, combined with the GPS map annotation function, according to the longitude and latitude information collected by the device, the geographical location of the device node is accurately marked and displayed on the map, enhancing the intuitiveness and practicality of device management;

[0065] The real-time acquisition interface realizes the real-time data acquisition function. The system can collect X / Y / Z three-component seismic data in real time and display the change trend of data acquisition through a dynamic curve graph. The acquisition results support visual display, allowing users to intuitively view the current data status and save the acquired data as a SEGY format file for subsequent data analysis. In addition, the system provides a file viewing function. By clicking on the file name, users can quickly obtain the basic information of the acquired data, including the sampling rate, number of samples, and timestamp, etc.;

[0066] The log interface realizes the log management function, which is used to record the system operation logs, device error logs, and user operation logs, facilitating users to maintain and manage the system. Through the error log viewing function, users can quickly troubleshoot common system failures such as network timeouts and file format errors. In addition, the system also provides a data processing interface, which displays the system resource usage and log statistics information through visual charts, supporting a comprehensive audit and analysis of user operation behaviors and system performance;

[0067] In addition, the front-end layer also includes a manager supervision interface, which includes the visual statistical analysis functions of log information, user activity distribution, system resource occupancy, and API request volume. Through data visualization auditing, managers can quickly analyze key information such as user active time, system performance status, and operation behavior distribution. In addition, the system has completed the visual display of each module, including user activity statistics, API request volume analysis, and performance monitoring, helping managers efficiently maintain the stability and reliability of system operation;

[0068] The back-end layer is developed based on the Spring Boot framework, which includes a device communication module, a log and monitoring module, a message queue module, a configuration management module, a data processing module, and a cache management module. The device communication module is responsible for accessing device data and completing data parsing and status feedback; the configuration management module supports the maintenance of system configuration parameters, network topology, and historical configurations; the data processing module is responsible for parsing and storing the acquired seismic data; the log and monitoring module realizes the recording of system error logs and user operation logs and the monitoring of the running status; the cache management module uses Redis to provide an efficient data caching service; the message queue module realizes asynchronous message passing through RabbitMQ to improve task processing efficiency;

[0069] The database layer mainly uses a MySQL relational database, which is responsible for storing device information, collected data, system configurations, user operation records, and log information. It includes device information tables, collected data tables, configuration management tables, log record tables, and operation record tables to ensure efficient data query and management. The database supports querying and exporting historical data to ensure the integrity and traceability of data during long-term storage and use. Through reasonable index design and backup strategies, the database ensures the performance and security of data in high-concurrency access scenarios;

[0070] The distributed deployment layer is responsible for the load balancing of system services, supports data synchronization and status monitoring, and provides reliable operation guarantees for backend services and data storage. It includes Docker containerization and a Spring Boot microservices architecture. Through the splitting of the Spring Boot microservices architecture, each functional module of the system can be independently deployed and managed, enhancing the scalability and stability of the system;

[0071] In this embodiment, by combining the SDN controller with the fusion routing algorithm, it can dynamically adapt to network state changes, achieve the purpose of dynamically adjusting and optimizing the routing path. Through dynamic path optimization, packet loss and retransmission are reduced, node energy consumption is balanced, local overload is avoided, thereby improving data transmission efficiency and meeting usage requirements.

[0072] In this embodiment, the unified state awareness and preprocessing module collects real-time network global state information through the global view function of SDN, and the collaborative control intelligent decision-making module optimizes the weight parameters according to the network global state information to ensure that the routing selection takes into account energy consumption balance, load distribution, and transmission delay. At the same time, the global path optimization targets high-priority tasks to generate the optimal path to ensure high reliability and low latency of transmission. The local path adjustment quickly provides a path correction plan for network dynamic changes and dynamically allocates path resources according to the priority of tasks. High-priority tasks preferentially select the global path, while ordinary tasks mainly rely on the local path, and an integrated routing strategy of global and local is output. Then, the control strategy generation and deployment module, according to the integrated routing strategy generated by the collaborative control intelligent decision-making module, uses the DQL-HRA routing algorithm module to perform network load balancing and energy consumption optimization from a global perspective, and uses the reinforcement learning algorithm to dynamically perceive the local network state and perform real-time path adjustment, converts the path selection scheme into a specific flow table, and quickly distributes the generated flow table to each forwarding node in the network through the SDN controller to guide the efficient forwarding and transmission execution of data traffic, ensuring the efficient transmission of different types of tasks and the reasonable utilization of resources.

[0073] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

Claims

1. A seismic wave acquisition system based on SDN and a fusion routing algorithm, characterized in that It includes an SDA controller, a fusion routing intelligent closed-loop control framework, a DQL-HRA routing algorithm module, and a software architecture. The fusion routing intelligent closed-loop control framework is deployed in the SDA controller. The fusion routing intelligent closed-loop control framework has a built-in fusion routing algorithm module. The fusion routing intelligent closed-loop control framework includes a unified state awareness and preprocessing module, a collaborative control intelligent decision-making module, and a control strategy generation and deployment module; The DQL-HRA routing algorithm module includes a centralized load balancing reward function and a distributed load balancing reward function; The software architecture includes a business layer, a front-end layer, a back-end layer, a database layer, and a distributed deployment layer.

2. The seismic wave acquisition system based on SDN and the fusion routing algorithm according to claim 1, wherein The unified state awareness and preprocessing module has functions of global state awareness, data preprocessing, and policy feedback and update. Through the global view function of SDN, it collects real-time network global state information and senses and summarizes the state information from forwarding nodes. At the same time, the unified state awareness and preprocessing module preprocesses the collected data to provide high-quality input data for subsequent modules, improving the accuracy and reliability of decision-making. The unified state awareness and preprocessing module analyzes the transmission results of completed tasks in a timely manner to optimize and update the next-round routing policy, realizing closed-loop control and dynamic adjustment; The network global state information includes network node load, link quality, and remaining energy data.

3. The seismic wave acquisition system based on SDN and the fusion routing algorithm according to claim 1, wherein The collaborative control intelligent decision-making module has functions of dynamic optimization of the reward function, global and local path collaborative decision-making, and task priority management. Combining reinforcement learning and collaborative control functions, it dynamically adjusts the weights of the centralized and distributed reward functions to meet the multi-objective optimization requirements. At the same time, it can achieve global path optimization and local path adjustment. Using the reinforcement learning algorithm, it optimizes the weight parameters according to the real-time network state to ensure that the routing selection takes into account the goals of energy consumption balance, load distribution, and transmission delay. Global path optimization is for high-priority tasks to generate the optimal path to ensure high reliability and low latency of transmission; local path adjustment is for network dynamic changes to quickly provide path correction solutions, dynamically allocating path resources according to the priority of tasks. High-priority tasks preferentially select the global path, and ordinary tasks mainly rely on the local path. Finally, the collaborative control intelligent decision-making module outputs the comprehensive routing policy of global and local for subsequent modules to execute and deploy.

4. The seismic wave acquisition system based on SDN and the fusion routing algorithm according to claim 1, characterized in that, The control strategy generation and deployment module has functions of routing strategy generation and flow table deployment. According to the comprehensive routing policy generated by the collaborative control intelligent decision-making module driven by reinforcement learning, it converts the path selection scheme into a specific flow table. The flow table includes the scheduling rules of the global path and the local path, supports the information annotation of task priorities, and ensures that high-priority tasks are preferentially guaranteed. Through the policy deployment interface of SDN, this module quickly distributes the generated flow table to each forwarding node in the network to guide the efficient forwarding and transmission execution of data traffic, ensuring the efficient transmission of different types of tasks and the reasonable utilization of resources.

5. The seismic wave acquisition system based on SDN and the fusion routing algorithm according to claim 1, characterized in that, The centralized load balancing reward function is used to achieve network energy consumption balance, load balance, minimization of transmission delay, and path selection stability. Its function formula is as follows: R global = λ1·Energy_Balance - λ2·Load_Variance - λ3·Delay + λ4·R history ; where R global is the global reward function, Energy_Balance is the energy balance, Load_Variance is the load variance, Delay is the path transmission delay, and R history is the historical reward smoothing term; The energy balance reflects the degree of balance in the distribution of the remaining energy of each node on the path. By measuring the balance of energy consumption of nodes on the path, this metric guides the routing selection to favor nodes with more remaining energy, thereby dispersing the overall energy consumption of the network, avoiding premature failure of a certain node due to energy exhaustion, and extending the network lifetime. The formula it uses is as follows: where |P| is the number of nodes on the path, is the remaining energy of node i, is the initial energy of node i; The load variance is used to measure the balance of the loads of each node in the network. The smaller the load variance, the more uniform the load distribution in the network, effectively reducing the data transmission bottleneck caused by an overloaded single node, and at the same time improving the stability and reliability of network transmission. The formula it uses is as follows: where L i is the current load of node i, is the average load of all nodes in the network, and |N| is the total number of nodes in the network; The path transmission delay reflects the cumulative transmission delay of data packets from the source node to the target node. The path transmission delay is one of the key metrics for evaluating network performance. Especially for high-priority tasks, a lower path transmission delay can ensure the real-time nature and transmission efficiency of data, helping to meet the requirements of delay-sensitive applications. The formula it uses is as follows: Among them, d j is the transmission delay of link j on the path; The historical reward smoothing term is used to smooth the dynamic changes between the current reward and the historical reward. Introducing the historical reward smoothing term can effectively suppress the frequent changes in path selection caused by short-term reward fluctuations, improve the stability of routing decisions, and ensure that the path planning can achieve a smooth transition under the long-term optimization goal. The formula it uses is: where k is the time window size of the historical reward, and R t represents the reward value at time t.

6. The seismic wave acquisition system based on SDN and the fusion routing algorithm according to claim 1, wherein The distributed load balancing reward function depends on the local information of the node, dynamically evaluates the load status, energy consumption level, and link quality of neighbor nodes in real time, is applicable to data transmission and path adjustment of ordinary tasks, and can quickly adapt to the dynamic changes of the network. The formula it uses is as follows: R local = μ1·ICATT i + μ2·Energy i - μ3·Load i - μ4·PLR i + μ5·R history ; Among them, ICATT i is the link load and interference level, Energy i is the energy consumption status of the node, Load i is the load status of the node, PLR i is the link packet loss rate; ICATT i By comprehensively considering the load of neighbor nodes and the link interference situation to measure the load pressure and interference state of the current path, and by optimizing the load and interference situation of the link, the transmission bottleneck can be effectively reduced and the data transmission quality can be improved. The formula it uses is as follows: Where N i is the set of neighbor nodes of node i, L j is the load of neighbor node j, T j is the transmission rate of neighbor node, NL j is the length of the send queue of neighbor node; Energy i It is used to measure the proportion of the remaining energy of a node in the total energy. This parameter ensures that nodes with high remaining energy are preferentially selected, which helps to evenly distribute the overall energy consumption of the network and extend the network's lifespan. The formula it uses is: where is the remaining energy of node i, is the initial energy of node i; Load i It is used to measure the load status of nodes. The greater the node load, the lower the reward obtained. By dynamically evaluating the load of nodes, the overload phenomenon of high-load nodes can be effectively avoided, thereby improving the stability and balance of the network. The formula it uses is as follows: Where L i is the current load of the node, T i is the actual transmission rate of the node, and NL i represents the queue length of the node's sending interface; PLR i It is used to measure the reliability of the current link. The higher the packet loss rate, the greater the penalty for rewards. By reducing the packet loss rate, the reliability and success rate of data transmission can be effectively improved, ensuring the stable operation of network performance. The formula it uses is:

7. The seismic wave acquisition system based on SDN and the fusion routing algorithm according to claim 1, characterized in that, The service layer includes a command control module, a response module, a status monitoring module, a data display module, and a data storage module. Among them, the command control module is responsible for managing and distributing the operation requirements of users; the response module processes the data interaction between the device and the system; the status monitoring module realizes the real-time detection of the device operation status; the data display module provides data visualization display; The data storage module ensures the secure storage of historical data and log information; The front-end layer is developed based on Web technology. It includes a data display interface, a system settings interface, a device management interface, a real-time acquisition interface, and a log interface. The front-end layer not only displays the results of business logic processing, but also can receive user input, complete system configuration and instruction issuance through interface operations. Moreover, the data displayed by the front-end layer is dynamically updated through a unified data interface to ensure the real-time and accuracy of the data; The back-end layer is developed based on the Spring Boot framework. It includes a device communication module, a log and monitoring module, a message queue module, a configuration management module, a data processing module, and a cache management module. The device communication module is responsible for accessing device data, completing data parsing and status feedback; the configuration management module supports the maintenance of system configuration parameters, network topology, and historical configurations; The data processing module is responsible for parsing and storing the collected seismic data; the log and monitoring module realizes the recording of system error logs, user operation logs, and operation status monitoring; the cache management module uses Redis to provide an efficient data caching service; the message queue module realizes asynchronous message passing through RabbitMQ to improve the task processing efficiency.

8. The seismic wave acquisition system based on SDN and the fusion routing algorithm according to claim 1, characterized in that The database layer mainly uses a MySQL relational database, which is responsible for storing device information, collected data, system configurations, user operation records, and log information. It includes a device information table, a data collection table, a configuration management table, a log record table, and an operation record table to ensure efficient query and management of data. The database supports querying and exporting of historical data, ensuring the integrity and traceability of data during long-term storage and use. Through reasonable index design and backup strategies, the database guarantees the performance and security of data in high-concurrency access scenarios; The distributed deployment layer is responsible for load balancing of system services, supports data synchronization and status monitoring, and provides reliable operation guarantees for backend services and data storage. It includes Docker containerization and a Spring Boot microservices architecture. Through the splitting of the Spring Boot microservices architecture, each functional module of the system can be independently deployed and managed, enhancing the scalability and stability of the system.

Images