An AI-based emergency communication link intelligent diagnosis and dynamic optimization method

Abstract

This invention discloses an AI-based intelligent diagnosis and dynamic optimization method for emergency communication links. Employing a system comprised of a dual-machine hot-standby core server, base station, relay station, and terminal equipment, it achieves stable links, continuous voice communication, and timely commands in emergency rescue scenarios. This method uses AI to provide intelligent diagnosis and dynamic optimization solutions, ensuring stable rescue communications. The method enables a real-time link monitoring module to collect parameter data at 20ms intervals; edge computing reduces bandwidth usage by ≥30%; dual-backup transmission is implemented; the AI ​​intelligent diagnosis module achieves a test set recognition accuracy of ≥95%; the dynamic link optimization module calls a strategy library to generate commands, supporting service priority scheduling; and the emergency command and coordination module provides visual display, including multi-level early warnings. Using this system improves the accuracy of anomaly diagnosis and shortens link interruption recovery time.

Description

Technical Field

[0001] This invention relates to the field of emergency communications, and more specifically, to an AI-based intelligent diagnosis and dynamic optimization method for emergency communication links. Background Technology

[0002] In emergency scenarios, the stable and efficient operation of communication links is crucial, directly impacting key aspects such as the transmission of rescue instructions, reporting of disaster situations, and the allocation of rescue resources. However, current emergency communication links suffer from numerous problems. On one hand, the emergency communication environment is complex and variable, potentially affected by natural factors such as terrain and weather, as well as human or technical factors such as equipment failure and signal interference, leading to unstable communication link performance and susceptibility to congestion and interruptions. On the other hand, existing methods for diagnosing and optimizing emergency communication links largely rely on manual operation or simple rule-based judgments, lacking intelligence and automation. This makes it difficult to quickly and accurately identify link faults and performance bottlenecks, and also hinders dynamic optimization and resource allocation based on real-time conditions, severely impacting the efficiency and effectiveness of emergency response.

[0003] For example, after major disasters such as earthquakes and floods, ground communication infrastructure may be severely damaged, and communication links may frequently fail. Traditional diagnostic methods require technicians to go to the site for inspection, which is not only time-consuming and labor-intensive, but may also prevent timely work due to the dangers of the disaster site. At the same time, when communication resources are scarce, traditional resource allocation methods are unable to allocate resources reasonably according to the priority of different rescue tasks, resulting in obstruction of critical information transmission and affecting the smooth progress of rescue work. Summary of the Invention

[0004] In view of the defects and problems of the existing technology, the purpose of this invention is to provide an AI-based intelligent diagnosis and dynamic optimization method for emergency communication links.

[0005] The technical solution adopted in this invention is: an AI-based intelligent diagnosis and dynamic optimization method for emergency communication links. This method employs a system consisting of a dual-machine hot-standby core server, a base station, a relay station, and terminal equipment. In emergency communication scenarios, it achieves stable links, continuous voice communication, and timely instructions for emergency rescue. The method steps are as follows:

[0006] The first, second, and third link real-time monitoring modules contained in S1, base stations, relay stations, and terminal equipment respectively send the collected valid data to the AI ​​intelligent diagnostic module of the dual-machine hot standby core server.

[0007] S2, the AI ​​intelligent diagnostic module will perform CRC32 reverse verification and forward verification on the received valid data. If the verification passes, wavelet denoising and Min-Max standardization processing will be performed; otherwise, the preset mechanism sub-process will be entered.

[0008] The S3 AI intelligent diagnostic module improves the feature extraction of the transformer model architecture through a sliding window, performs anomaly judgment, type identification, location positioning and level assessment, saves the obtained and confirmed diagnostic results to the diagnostic data storage device, and outputs them to the dynamic link optimization module and emergency command and coordination module of the dual-machine hot standby core server.

[0009] S4. The dynamic link optimization module will receive the diagnostic results, generate optimization instructions, send them to the base station, then send them to the relay station, and finally send them to the terminal device for execution.

[0010] After receiving the diagnostic results, the S5 emergency command and coordination module updates the equipment anomaly report and optimization log, and pushes the data to the base station's emergency management platform through the AI ​​network management platform of the dual-machine hot standby core server to synchronize the data.

[0011] The dual-machine hot standby core server includes an AI network management platform, an AI intelligent diagnostic module, a diagnostic data storage device, an emergency command and coordination module, and a dynamic link optimization module; the base station includes a first link real-time monitoring module, an optical fiber interface card, and an emergency management platform; the relay station includes a second link real-time monitoring module and a first 5G uRLLC module; and the terminal equipment includes a third link real-time monitoring module and a second 5G uRLLC module.

[0012] The beneficial effects of this invention are as follows: This method uses AI to provide intelligent diagnosis and dynamic optimization solutions, ensuring stable rescue communications. Using this method, the link real-time monitoring module collects parameter data at 20ms intervals; edge computing reduces bandwidth usage by ≥30%; dual-backup transmission is implemented; the AI ​​intelligent diagnosis module achieves a test set recognition accuracy of ≥95%; the dynamic link optimization module calls the policy library to generate instructions, supporting service priority scheduling; the emergency command and coordination module provides visual display, including multi-level early warnings; using this system can improve the accuracy of anomaly diagnosis and shorten link interruption recovery time. Attached Figure Description

[0013] Figure 1 This is a diagram showing the overall system framework of the present invention;

[0014] Figure 2 This is a flowchart illustrating the steps of an embodiment of the present invention;

[0015] Figure 3 This is a flowchart of the preset mechanism in an embodiment of the present invention. Detailed Implementation

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] Reference Figure 1 , Figure 2The AI-based intelligent diagnosis and dynamic optimization method for emergency communication links adopts a system consisting of a dual-machine hot standby core server, base station, relay station and terminal equipment. In emergency communication scenarios, it can achieve the functions of link stability, voice continuity and timely instructions in emergency rescue.

[0018] The dual-machine hot standby core server contains an AI network management platform, an AI intelligent diagnostic module, a diagnostic data storage device, an emergency command and coordination module, and a dynamic link optimization module; the base station contains a first link real-time monitoring module, an optical fiber interface card, and an emergency management platform; the relay station contains a second link real-time monitoring module and a first 5G uRLLC module; and the terminal equipment contains a third link real-time monitoring module and a second 5G uRLLC module.

[0019] The steps of the AI-based intelligent diagnosis and dynamic optimization method for emergency communication links are as follows:

[0020] The first, second, and third link real-time monitoring modules contained in S1, base stations, relay stations, and terminal equipment respectively send the collected valid data to the AI ​​intelligent diagnostic module of the dual-machine hot standby core server.

[0021] The first, second, and third link real-time monitoring modules are deployed at the core nodes (base stations, relay stations) and terminal equipment (multi-mode emergency walkie-talkies, emergency command terminals) of the emergency communication network, respectively. They require signal acquisition hardware support and have the function of acquiring signal strength, bit error rate, data packets, transmission delay, and signal-to-noise ratio.

[0022] The first, second, and third link real-time monitoring modules preprocess the collected valid data, perform local filtering and outlier removal, and transmit the preprocessed valid data to the AI ​​intelligent diagnostic module through the low-latency channel formed by the fiber optic interface card in the base station and the first and second 5G uRLLC modules in the relay station and terminal equipment, respectively. The local preprocessing latency is ≤10ms, reducing bandwidth usage by ≥30%.

[0023] The first, second, and third link real-time monitoring modules collect effective link parameter data in real time at a collection cycle of 20ms / time. The parameters include physical layer signal strength (unit: dBm), bandwidth utilization (unit: %), bit error rate (unit: %), transmission delay (unit: ms), signal-to-noise ratio (unit: dB), and link interruption frequency (unit: times / minute).

[0024] The low-latency channel adopts a 5G uRLLC + fiber optic dual-backup architecture. Terminal devices and relay stations transmit data to the relay station and base station respectively through their own 5G uRLLC modules. The base station transmits data to the AI ​​intelligent diagnostic module through a fiber optic interface card. Data integrity is ensured through CRC32 verification before transmission, and TLS 1.3 encryption is used to prevent data leakage.

[0025] S2, the AI ​​intelligent diagnostic module will perform CRC32 reverse verification and forward verification on the received valid data. If the verification passes, wavelet denoising and Min-Max standardization processing will be performed; otherwise, the preset mechanism sub-process will be entered.

[0026] The AI ​​intelligent diagnostic module is included in the dual-machine hot standby core server. It is connected to the PCIe 4.0 slot on the motherboard of the dual-machine hot standby core server via a PCIe 4.0 x16 interface and includes a diagnostic data storage device.

[0027] The AI ​​intelligent diagnostic module first confirms that the received valid data has not been tampered with or lost through CRC32 reverse verification. Then, it performs forward verification on the CRC32 reverse verification diagnostic result by comparing the deviation threshold between real-time data and historical normal data. If the verification consistency is ≥95%, the diagnostic result is output. If the verification consistency is <95%, the diagnosis is re-diagnosed and the preset mechanism sub-process is entered.

[0028] like Figure 3 As shown, the preset mechanism sub-process performs the following operations:

[0029] The AI ​​intelligent diagnostic module sends a re-acquisition command to the corresponding link real-time monitoring module. The link real-time monitoring module shortens the acquisition cycle, increases the sample size, and performs a second verification. If the verification consistency is ≥95%, the second verification passes, and the preset mechanism ends. If the verification consistency is still <95%, the second verification fails. A third verification is then performed after adding a median filtering step and optimizing the Min-Max standardized numerical range. If the verification consistency is ≥95%, the third verification passes, and the preset mechanism ends. If the verification consistency is still <95%, the third verification fails. An orange alert is triggered by the emergency command and coordination module. The dynamic link optimization module immediately generates and issues a backup link switching command, calls a lightweight CNN model, extracts signal strength, transmission latency, and bit error rate, and restricts non-core tasks.

[0030] The AI ​​intelligent diagnostic module sends a re-acquisition command to the corresponding link real-time monitoring module through a low-latency channel. The link real-time monitoring module shortens the default acquisition cycle of 20ms to 10ms, continuously acquires three sets of data, and merges and transmits them. If the consistency of the second verification is less than 95%, the link real-time monitoring module starts enhanced preprocessing, adds a median filtering step, and the AI ​​intelligent diagnostic module simultaneously optimizes the Min-Max normalization range to [-1,1]. If the consistency of the third verification is still less than 95%, the emergency command and coordination module immediately triggers an orange alert, and the alert information is synchronized to the AI ​​network management platform and the emergency management platform. The dynamic link optimization module issues a "backup link switch" command via the UDP protocol (latency ≤ 50ms), and the base station, relay station, and terminal equipment switch to the pre-configured backup link in a coordinated manner. The AI ​​intelligent diagnosis module calls a lightweight CNN diagnostic model to extract only three core parameter features: signal strength, transmission latency, and bit error rate, and quickly outputs simplified diagnostic results, ensuring that the diagnostic latency is ≤ 20ms. The dynamic link optimization module simultaneously issues a "non-core business restriction" command to suspend the transmission of non-emergency data at level 3 and limit the bandwidth occupied by emergency commands at level 2 to ≤ 20%, so as to fully guarantee the continuity of the level 1 voice intercom service.

[0031] The S3 AI intelligent diagnostic module improves the feature extraction of the transformer model architecture through a sliding window, performs anomaly judgment, type identification, location positioning and level assessment, saves the obtained and confirmed diagnostic results to the diagnostic data storage device, and outputs them to the dynamic link optimization module and emergency command and coordination module of the dual-machine hot standby core server.

[0032] The AI-powered intelligent diagnostic module extracts time-series data through a sliding window, extracts time-domain-frequency domain fusion features through a pre-trained model, and completes anomaly detection, type identification, location positioning, and severity assessment based on the features. The diagnostic results are confirmed through a multi-model fusion verification mechanism, and the diagnostic results information is finally saved to the diagnostic data storage device and output to the dynamic link optimization module and the emergency command and coordination module.

[0033] S4. The dynamic link optimization module receives the diagnostic results, calls the pre-stored optimization strategy library (containing optimization templates for different scenarios) to generate optimization instructions, sends them to the base station, then to the relay station, and finally to the terminal device for execution. The instructions include: authorized emergency frequency band switching, adaptive adjustment of transmit power, multi-link load balancing, and backup link switching (pre-configured with 2-3 backup links).

[0034] The dynamic link optimization module is deployed in the dual-machine hot standby core server. Based on the diagnostic results, it matches the following strategy library: 1. Signal attenuation (general level): Generates an adaptive adjustment command for transmit power; 2. Frequency band interference (general level): Generates an authorized emergency frequency band switching command; 3. Link congestion (important level): Generates a multi-link load balancing command; 4. Link interruption (emergency level): Generates a backup link switching command. After the command is generated, it is sent to the base station via UDP protocol (latency ≤ 50ms).

[0035] After receiving the diagnostic results, the S5 emergency command and coordination module updates the equipment anomaly report and optimization log, and pushes the data to the base station's emergency management platform through the AI ​​network management platform of the dual-machine hot standby core server (displaying link topology, anomaly annotation, and optimization curve) to synchronize the data.

[0036] The emergency command and coordination module has a multi-level early warning function: Level 3 anomaly triggers a yellow warning, Level 4 triggers an orange warning, and Level 5 triggers a red warning, with a warning response delay of ≤50ms.

[0037] Example: In an emergency communication scenario, the system utilizes standard emergency communication equipment (base stations, relay stations, and terminal devices), all of which are wirelessly connected. The dual-machine hot standby core server is connected to the base station via a wired connection.

[0038] Emergency responders use terminal devices to maintain normal business operations in areas without base station signal coverage via relay stations. In this scenario, the real-time link monitoring modules in the base station, relay station, and terminal devices collect link parameters in real time and send them to the AI ​​intelligent diagnostic module through a low-latency channel.

[0039] After receiving data from the link real-time monitoring module, the AI ​​intelligent diagnosis module first verifies the data integrity through CRC32 reverse verification, then performs noise reduction, standardization, and time series data extraction. Next, it calls the pre-trained model to extract features and complete the verification. Finally, it verifies the consistency of the diagnosis results through multi-model fusion and outputs the data to the dynamic link optimization module and the emergency command and coordination module. If the diagnosis results are inconsistent, they are handled according to the preset mechanism.

[0040] The emergency command and coordination module receives diagnostic results and displays the link topology, anomaly markers, and optimization curves through the AI ​​network management platform, while simultaneously synchronizing the data to the emergency management platform. In the event of a link failure, the emergency command and coordination module will select SMS or voice alarm based on the link failure level and send alarm commands to the AI ​​network management platform.

[0041] After receiving the diagnostic results, the dynamic link optimization module calls the pre-stored strategy library, matches the corresponding optimization template, generates instructions, and sends the instructions to the base station via the UDP protocol. The base station then performs the corresponding operation based on the instructions.

[0042] By implementing this method and system, the accuracy of emergency communication link anomaly diagnosis is improved, and the link interruption recovery time is shortened compared to manual handling, thus meeting the needs of stable emergency rescue links, continuous voice communication, and timely instructions.

[0043] In this embodiment, the pre-training dataset of the AI ​​intelligent diagnosis module is divided into training set, validation set and test set in an 8:1:1 ratio. The accuracy rate of manually annotated abnormal samples is ≥98%, and the model's anomaly recognition accuracy on the test set is ≥95% with a false positive rate of ≤3%.

[0044] In this embodiment, the dynamic link optimization module adopts service priority scheduling: emergency services are divided into three levels (level 1 voice intercom, level 2 emergency commands, and level 3 non-emergency data). When bandwidth is insufficient, the usage of level 2 and level 3 services is limited (level 2 ≤ 30%, level 3 ≤ 20%), ensuring that the packet loss rate of level 1 services is ≤ 1% and the latency is ≤ 200ms.

[0045] The adaptive protocol supports mainstream emergency response standards such as PDT and achieves cross-standard parameter interaction through protocol adaptation algorithms, with an adaptation latency of ≤30ms.

[0046] The system adopts a dual-machine hot standby core server, with the two core devices synchronizing data every 10ms. This enables the standby device to take over within 50ms when the main device fails, ensuring a fault recovery time of ≤3 minutes and guaranteeing continuous system operation.

[0047] The dual-machine hot standby core server consists of two core servers with identical functions. The primary server provides business services normally, while the standby server synchronizes the data and operating status of the primary server in real time and is in standby mode.

Claims

1. An AI-based intelligent diagnosis and dynamic optimization method for emergency communication links, characterized in that, A system consisting of a dual-machine hot standby core server, base station, relay station, and terminal equipment is adopted to achieve stable links, continuous voice communication, and timely instructions in emergency communication scenarios. The method and steps are as follows: S1, the base station, relay station and terminal equipment respectively contain the first, second and third link real-time monitoring modules, which send the collected valid data to the AI ​​intelligent diagnosis module of the dual-machine hot standby core server; S2, the AI ​​intelligent diagnosis module will perform CRC32 reverse verification and forward verification on the received valid data. If the verification is successful, wavelet denoising and Min-Max standardization processing will be performed; otherwise, the preset mechanism sub-process will be entered. The S3 AI intelligent diagnosis module captures time-series data through a sliding window, extracts features from the transformer pre-trained model architecture, performs anomaly judgment, type identification, location positioning, and level assessment, saves the obtained and confirmed diagnosis results to the diagnosis data storage device, and outputs them to the dynamic link optimization module and emergency command and coordination module of the dual-machine hot standby core server. S4. The dynamic link optimization module will receive the diagnostic results, generate optimization instructions, send them to the base station, then send them to the relay station through the base station, and finally send them to the terminal device to execute the instructions. After receiving the diagnostic results, the S5 emergency command and coordination module updates the equipment anomaly report and optimization log, and pushes the data to the base station's emergency management platform through the AI ​​network management platform of the dual-machine hot standby core server to synchronize the data. The preset mechanism sub-process performs the following operations: The AI ​​intelligent diagnostic module sends a re-collection command to the corresponding link real-time monitoring module. The link real-time monitoring module shortens the collection cycle, increases the number of collected samples, and performs a second verification. If the verification consistency is ≥95%, the second verification passes, and the preset mechanism ends. If the consistency of the verification is still <95%, the second verification fails. A third verification is then performed after adding a median filtering step and optimizing the Min-Max standardized numerical range. If the consistency of the verification is ≥95%, the third verification passes and the preset mechanism ends. If the consistency of the verification is still <95%, the third verification fails. The emergency command and coordination module triggers an orange alert, and the dynamic link optimization module immediately generates and issues a backup link switching instruction, calls a lightweight CNN model to extract signal strength, transmission delay, and bit error rate, and restricts non-core tasks.

2. The AI-based intelligent diagnosis and dynamic optimization method for emergency communication links according to claim 1, characterized in that, The first, second, and third link real-time monitoring modules preprocess the collected valid data, perform local filtering and outlier removal, and transmit the preprocessed valid data to the AI ​​intelligent diagnostic module through the low-latency channel formed by the fiber optic interface card in the base station and the first and second 5G URLLC modules in the relay station and terminal equipment, respectively. The local preprocessing latency is ≤10ms, reducing bandwidth usage by ≥30%.

3. The AI-based intelligent diagnosis and dynamic optimization method for emergency communication links according to claim 2, characterized in that, The AI ​​intelligent diagnostic module first confirms that the received valid data has not been tampered with or lost through CRC32 reverse verification. Then, it performs forward verification on the CRC32 reverse verification diagnostic result by comparing the deviation threshold between real-time data and historical normal data. If the verification consistency is ≥95%, the diagnostic result is output. If the verification consistency is <95%, the diagnosis is re-diagnosed and the preset mechanism sub-process is entered.

4. The AI-based intelligent diagnosis and dynamic optimization method for emergency communication links according to claim 3, characterized in that, The dynamic link optimization module receives the diagnostic results and calls the pre-stored optimization strategy library to generate optimization instructions, which include: authorized emergency frequency band switching, adaptive adjustment of transmit power, multi-link load balancing, and backup link switching.

5. The AI-based intelligent diagnosis and dynamic optimization method for emergency communication links according to claim 4, characterized in that, The emergency command and coordination module has a multi-level early warning function: Level 3 anomaly triggers a yellow warning, Level 4 triggers an orange warning, and Level 5 triggers a red warning, with a warning response delay of ≤50ms.

6. The AI-based intelligent diagnosis and dynamic optimization method for emergency communication links according to claim 1, characterized in that, The dual-machine hot standby core server includes an AI network management platform, an AI intelligent diagnostic module, a diagnostic data storage device, an emergency command and coordination module, and a dynamic link optimization module; the base station includes a first link real-time monitoring module, an optical fiber interface card, and an emergency management platform; the relay station includes a second link real-time monitoring module and a first 5G uRLLC module; and the terminal equipment includes a third link real-time monitoring module and a second 5G uRLLC module.

Images