An intelligent emergency early warning system and method based on multi-network fusion and edge computing

Abstract

The application discloses an intelligent emergency early warning system and method based on multi-network fusion and edge computing, and relates to the technical field of emergency management.The system is uniformly dispatched by a main control module through a bus to four modules of power supply, protection, communication and audio.The main control module performs cooperative control according to feedback signals of the modules, including power supply mode switching, image analysis, sensor data acquisition, communication intelligent optimization based on link state and SM4 encryption processing, and supports voice synthesis, noise reduction filtering and hierarchical early warning response.The power supply module integrates solar energy, a storage battery and a mains unit to realize dynamic power optimization and low-power core protection;the protection module provides integrated sealing, lightning protection and impact resistance;the communication module integrates 4G / 5G and L / S / Ku multi-band satellite communication to support link adaptive switching and redundancy backup;and the audio module drives a loudspeaker to complete sound amplification and broadcast after digital-to-analog conversion, amplification and filtering.

Description

Technical Field

[0001] This invention relates to the field of emergency management technology, and in particular to an intelligent emergency early warning system and method based on multi-network fusion and edge computing. Background Technology

[0002] In emergency rescue and disaster early warning scenarios, the stability of communication links directly determines the efficiency of early warning information transmission and the success rate of rescue. Existing emergency terminals mostly use a single communication method, such as supporting only 4G / 5G SMS or BeiDou short messages. When extreme weather (such as rainstorms, typhoons, and earthquakes) causes ground communication base stations to fail or signals to be interrupted, the terminals cannot receive and transmit early warning information normally, resulting in communication blind spots. Some multi-communication terminals require manual switching of communication methods, which is cumbersome, inefficient in emergency response, and cannot meet the rapid response needs in emergency scenarios.

[0003] Meanwhile, existing emergency terminals have relatively limited functions, mostly only capable of receiving information or simple broadcasting, lacking multi-format conversion of warning information (such as text, SMS, and voice-to-audio amplification). For remote areas, outdoor operations, and other densely populated or dispersed scenarios, the coverage of warning information transmission is limited. In terms of power supply, some terminals rely solely on mains power or batteries, and cannot continue to operate when mains power is interrupted or there is no power supply at the disaster site. In addition, existing terminals generally lack comprehensive protection functions and environmental perception capabilities, making them easily damaged in harsh outdoor environments (such as high temperatures, heavy rain, lightning, and dust). They also cannot collect and analyze on-site environmental data and image information in real time, which is detrimental to the command center's understanding of the actual situation on site and affects the scientific and timely nature of emergency decision-making.

[0004] Therefore, there is an urgent need for an emergency early warning method that integrates multiple communication methods, can automatically switch communication links, has comprehensive functions, stable power supply, excellent protection performance, and environmental awareness capabilities, in order to solve the problems of poor communication reliability, slow emergency response, single function, and weak adaptability in existing technologies. Summary of the Invention

[0005] This application provides an intelligent emergency early warning system based on multi-network convergence and edge computing, including: a power supply module, a protection module, a communication module, and an audio module. These modules are connected via a bus and are uniformly controlled and scheduled by a main control module. The main control module performs system-wide collaborative control based on feedback signals from each module. This collaborative control includes power supply mode switching, image analysis, sensor data acquisition, intelligent communication link optimization based on communication link status, and information processing and SM4 encryption based on the type of received information. Information processing includes speech synthesis, noise reduction filtering, and early warning analysis. Based on the early warning analysis results, corresponding early warning actions are triggered. The power supply module... The module includes a solar power supply unit, a battery power supply unit, and a mains charging unit, used to dynamically select the power source based on the mains power status and implement a low-power core protection strategy; the protection module includes shell protection, lightning protection, and shock protection, used to achieve integrated sealed protection of the terminal; the communication module includes a 4G / 5G communication unit, an L-band satellite communication unit, an S-band satellite communication unit, and a Ku-band satellite communication unit, used to perform multi-mode intelligent switching and redundancy backup based on the communication link status and signal strength; the audio module includes a digital-to-analog converter, a preamplifier, a power amplifier, a speaker, and a volume control unit, used to amplify and broadcast the received audio signals.

[0006] This application also provides an intelligent emergency early warning method based on multi-network fusion and edge computing. After the system is powered on and completes initialization, firmware configuration loading, full module self-test, multi-mode communication link attachment, and storage module initialization, it enters normal monitoring mode. Under normal monitoring, the main control module performs low-power scheduling, the image analysis module performs low-frequency frame sampling and lightweight motion detection, the sensor interface module periodically collects environmental data and performs validity verification and sliding window filtering, and simultaneously monitors the communication link status and power supply status. When the data exceeds the threshold, an early warning response is initiated. During the early warning response, high-frequency full-frame rate image acquisition is switched, an early warning analysis report with timestamp and location tag is generated, abnormal sensor data is integrated, encrypted with national cryptographic standards, and transmitted back through the optimal communication link. High-bandwidth data storage is initiated simultaneously. The early warning information is parsed and a graded early warning strategy is matched. Corresponding power amplification alarm and adaptive volume adjustment are executed. The power supply source is dynamically monitored, and a low-power core protection strategy is executed when the battery is low. Non-core power consumption is reduced to ensure communication and alarm core functions. After the early warning is lifted, normal low-power monitoring is restored, the link health is monitored in real time, and multi-mode communication seamless switching and redundant backup takeover are executed when the link fails.

[0007] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. This invention adopts a multi-network converged full-domain communication architecture, integrating 4G / 5G public network communication units with L / S / Ku multi-band satellite communication units. Through intelligent link optimization algorithm and multi-mode hot backup redundancy switching mechanism, it calculates link health based on multi-dimensional parameter weighted fusion, monitors link status in real time and dynamically switches the optimal transmission channel, solving the core pain points of traditional emergency early warning terminals such as public network base station paralysis, communication interruption in remote areas without signal, and link switching interruption and packet loss. It achieves full-domain coverage, high reliability and uninterrupted continuous transmission of emergency information.

[0008] 2. This invention integrates an edge AI image analysis module and a multi-type environmental sensor interface module. Through a hierarchical intelligent acquisition strategy driven by scene dynamic features, edge target recognition and risk assessment algorithms, combined with a multi-sensor multi-level threshold early warning mechanism and a hierarchical early warning response strategy with dual-parameter joint indexing, it achieves rapid capture of abnormal hazards, accurate assessment of risk levels, and adaptive triggering of corresponding early warning actions. It solves the problems of unbalanced power consumption and monitoring accuracy in traditional terminal acquisition, delayed early warning response, and insufficient matching degree between risk and action, and significantly improves the intelligence level, accuracy of assessment, and timeliness of response of emergency early warning.

[0009] 3. This invention designs a multi-mode integrated power supply system combining mains power, solar power, and battery, coupled with an integrated sealed protection, lightning protection, and shock resistance architecture. Through a dynamic optimization mechanism that prioritizes mains power as the highest power source, a smooth switching strategy for power supply paths under all operating conditions, and a strategy to ensure core functions under low power conditions, it solves the problems of poor adaptability to harsh outdoor environments, insufficient battery life in scenarios without mains power, and failure of core emergency functions under extreme low power conditions in traditional emergency terminals. This achieves system deployment flexibility, adaptability to complex environments, and long-lasting high reliability.

[0010] 4. This invention utilizes a national cryptographic hardware encryption engine integrated into the main control module to perform national cryptographic encryption and decryption, integrity verification, and signature verification on all types of information, including remote control commands, warning texts, firmware upgrade packages, and audio streams. Combined with a full-process log retention mechanism for warning events and environmental monitoring data, this invention solves the problems of easily tampered data transmission in traditional emergency terminals, security vulnerabilities in firmware upgrades, and lack of traceability for emergency events. It ensures the security, integrity, and full-process traceability of emergency data and commands, providing reliable data support for emergency command review and disaster tracing.

[0011] 5. This invention achieves unified bus-based control and collaborative scheduling of the entire system through the main control module. Based on the real-time feedback signals of each module, it performs multi-dimensional logical comprehensive judgment and adaptive response. Combined with an audio amplification mechanism that adaptively adjusts gain to environmental noise and a firmware security hot upgrade function verified by national cryptographic standards, it realizes dynamic allocation of system resources and intelligent balance between performance and power consumption. It solves the problems of poor coordination of isolated operation of modules in traditional terminals, poor broadcasting effect in complex environments, and weak autonomous fault tolerance, thus ensuring the efficient coordination, low power consumption, stability, and long-term autonomous operation of the intelligent emergency early warning system. Attached Figure Description

[0012] Figure 1 A schematic diagram of the structure of an intelligent emergency early warning system based on multi-network fusion and edge computing provided in an embodiment of this application; Figure 2 A flowchart of an intelligent emergency early warning method based on multi-network fusion and edge computing provided for embodiments of this application. Detailed Implementation

[0013] This application provides an intelligent emergency early warning system and method based on multi-network convergence and edge computing. Addressing the pain points of existing emergency early warning terminals, such as easy communication interruption under extreme disaster conditions, low efficiency of manual switching, limited functionality, and poor environmental adaptability, this application discloses an intelligent emergency early warning system and method based on multi-network convergence and edge computing. The system uses a main control module to uniformly schedule power supply, protection, communication, and audio modules, integrating 4G / 5G and L / S / Ku multi-band satellite communication units and achieving automatic seamless switching and redundant backup based on link health. An image analysis module performs hierarchical data acquisition and edge AI recognition and analysis, generating real-time early warning reports with spatiotemporal tags and transmitting them back to the command center. Simultaneously, a sensor interface module enables multi-level threshold triggering of early warnings based on environmental data. Furthermore, the system features intelligent complementary power supply from solar energy, batteries, and mains power, along with a low-power core protection strategy. It adopts an integrated sealed, lightning-proof, and shock-resistant design, and supports text-to-speech synthesis and adaptive volume adjustment, significantly improving communication robustness, response timeliness, and on-site perception capabilities in emergency scenarios.

[0014] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0015] like Figure 1The diagram shows the structure of an intelligent emergency early warning system based on multi-network fusion and edge computing provided in this application embodiment. This system includes a power supply module, a protection module, a communication module, and an audio module. These modules are connected via a bus and are uniformly controlled and scheduled by a main control module. The main control module uses a high-performance MCU chip and performs system collaborative control based on feedback signals from each module. This collaborative control includes power supply mode switching, image analysis, sensor data acquisition, intelligent communication link optimization based on communication link status, and information processing and SM4 encryption based on the type of received information. Information processing includes speech synthesis, noise reduction filtering, and early warning analysis. Based on the early warning analysis results, corresponding early warning actions are triggered. The power supply module includes a solar power supply unit, a battery power supply unit, and a mains charging unit. It dynamically optimizes the power source based on the mains power status and executes a low-power core protection strategy. The solar power supply unit uses a 200W monocrystalline silicon solar panel, paired with a solar controller, to convert solar energy into electrical energy to charge the battery and directly power various terminal modules, adapting to outdoor scenarios without mains power. The battery power supply unit uses a 12V / 100Ah lithium iron phosphate battery, with capacity expandable as needed. A full charge supports continuous operation for ≥72 hours (without solar supplementation, under normal operating conditions). It features overcharge, over-discharge, and overcurrent protection. The AC220V mains charging unit supports AC220V mains input and is equipped with a charging management chip, allowing direct charging of the battery and simultaneous power supply to all modules of the terminal. When both mains and solar power are available, mains power is prioritized to reduce solar panel losses. The protection module includes casing protection, lightning protection, and impact resistance, providing integrated sealed protection for the terminal. The terminal's outer shell is made of 304 stainless steel with an anodized finish, providing rust and corrosion resistance, making it suitable for harsh outdoor environments (such as coastal areas and chemical plants). Waterproof sealing rings are used at the shell connections to prevent dust and rainwater from entering the terminal and damaging electronic components. Lightning protection includes surge protectors (using SPD surge protection modules) at the power and communication interfaces, capable of withstanding ≥20kV lightning strikes to prevent damage from lightning through power and communication lines. The internal electronic components are secured with shock-resistant brackets, and the outer shell features a buffer structure, capable of withstanding ≥1...The system withstands a 5m freefall impact test, adapting to the bumpy and collision scenarios in emergency situations. The communication module includes a 4G / 5G communication unit, an L-band satellite communication unit, an S-band satellite communication unit, and a Ku-band satellite communication unit. These units perform intelligent multi-mode switching and redundancy backup based on communication link status and signal strength, ensuring uninterrupted communication in extreme scenarios. The 4G / 5G communication unit supports 4G / 5G full network compatibility, responsible for receiving and sending text and voice messages, adapting to conventional communication environments with high communication speed and low cost. The L-band satellite communication unit supports the emergency BeiDou messaging protocol (GB / T45936-2025), enabling communication in scenarios without terrestrial communication signals. (In cases of base station paralysis due to earthquakes or floods), it enables the reception and transmission of information, boasts long communication distances and strong anti-interference capabilities. The S-band satellite communication unit supports voice calls, text messages, and data transmission, covering the entire country and addressing the communication needs of remote areas, oceans, deserts, and other areas without terrestrial communication or BeiDou signal coverage. The Ku-band satellite communication unit supports high-throughput data transmission, providing real-time communication services and supporting voice, data, and image transmission. The audio module includes a digital-to-analog converter, preamplifier, power amplifier, speaker, and volume control unit, with an output power ≥50W. The speaker is a waterproof full-range speaker used for amplifying and broadcasting the received audio signals.

[0016] The system employs a dual storage design using SD cards and Flash memory. The SD card has a capacity of ≥32GB (expandable to 128GB) for storing image data, environmental monitoring data, early warning information, and analysis reports. The Flash memory has a capacity of ≥8GB for storing terminal firmware, speech synthesis libraries, and algorithm models. It supports cyclic overwrite storage and locking of critical data to prevent accidental deletion or overwriting of important disaster and early warning data, ensuring long-term automatic operation of the terminal. The module supports intelligent storage scheduling. Under normal monitoring conditions, it operates in a low-power mode, storing only critical data. During abnormal triggers and early warnings, it automatically increases the storage rate to ensure that frequently collected images and real-time disaster data are not lost. Simultaneously, it supports encrypted data transmission and cloud-based synchronous backup. Local storage ensures data retention during network outages, and automatic synchronization to the backend platform after network recovery achieves "no local data loss, traceable cloud access." This meets the data storage needs of emergency scenarios while intelligently balancing storage performance and device power consumption, providing comprehensive data support for emergency command review and disaster tracing.

[0017] In this embodiment, the present invention uses a high-performance MCU chip in the main control module to uniformly schedule the various modules via a bus. This not only enables basic collaborative control such as power supply mode switching, image analysis, and sensor data acquisition, but also intelligently selects the best communication link based on its status. It performs speech synthesis, noise reduction filtering, and early warning analysis on different types of information, encrypting them using the national standard SM4. Based on the early warning analysis results, it triggers corresponding early warning actions. This ensures both the collaborative and intelligent operation of the system, while also guaranteeing the security of emergency information transmission and processing through national standard encryption technology, meeting the needs for efficient processing of diverse information in emergency scenarios. The power supply module integrates a 200W single... The system comprises a crystalline silicon solar power supply unit, a 12V / 100Ah expandable lithium iron phosphate battery power supply unit, and an AC220V mains charging unit. It dynamically selects the optimal power source based on the mains power status, prioritizing mains power when both mains and solar power are available to reduce solar panel losses. It also features a low-power core protection strategy, allowing the terminal to operate continuously for over 72 hours when the battery is fully charged and there is no solar supplementation. Furthermore, it includes overcharge, over-discharge, and overcurrent protection, making it suitable for emergency scenarios without mains power outdoors. It also completely solves the problems of unstable power supply and power failure in extreme environments, ensuring long-term continuous system operation. The protection module uses a 304 stainless steel anodized shell for protection. Rust and corrosion resistant design suitable for harsh outdoor environments such as seaside and chemical industrial zones. Combined with a waterproof sealing ring to block dust and rainwater, and an SPD lightning protection module capable of withstanding 20kV lightning strikes to prevent damage, the internal shock-absorbing bracket and outer shell buffer structure can withstand a 1.5m free-fall impact. Integrated sealing protection significantly improves the terminal's environmental adaptability and physical damage resistance, enabling it to easily cope with bumps, collisions, lightning, rain, snow, and other harsh conditions at emergency sites. The communication module integrates 4G / 5G and L / S / Ku multi-band satellite communication units. The 4G / 5G unit adapts to conventional scenarios to achieve high-speed, low-cost communication, while the L-band satellite unit relies on the emergency BeiDou messaging protocol to resolve grounding issues. To address communication challenges in scenarios where base stations are paralyzed, such as earthquakes and floods, the S-band satellite unit covers remote, ocean, and desert areas without ground or BeiDou signals, while the Ku-band satellite unit enables real-time transmission of high-bandwidth audio, video, and images. Multi-mode intelligent switching and redundant backup mechanisms completely solve the pain points of traditional emergency equipment's single communication mode and communication interruptions under extreme disasters, comprehensively ensuring uninterrupted communication in various extreme emergency scenarios. The audio module, through a digital-to-analog converter, multi-stage power amplifier, and a waterproof full-range speaker with an output power ≥50W, can amplify and broadcast received audio signals. Its waterproof characteristics make it suitable for outdoor environments, and its high-power output ensures clear transmission of warning information even in noisy emergency situations.

[0018] In addition, the system also includes an image analysis module and a sensor interface module. The image analysis module is used to perform hierarchical intelligent acquisition, edge AI recognition and analysis based on scene status and on-site image content, and generate an early warning analysis report to be sent back to the command center. The specific steps are as follows: S201, the image sensor configured in the image analysis module continuously captures the real-time image stream of the monitoring area, and performs preprocessing operations on the continuous frame images. The preprocessing operations include image denoising, illumination compensation and size normalization; S202, based on the scene dynamic feature parameters of the preprocessed image, weighted summation processing is performed to obtain the scene dynamic comprehensive score value. The scene type is dynamically determined based on the scene dynamic comprehensive score value. The scene dynamic feature parameter package... The dynamic comprehensive score of a scene is determined by the following parameters: inter-frame differential amplitude, motion vector distribution entropy, and regional pixel change rate. If the scene's dynamic comprehensive score is lower than a preset static threshold, it is classified as a static normal scene. If the scene's dynamic comprehensive score is not lower than the preset static threshold, it is classified as a dynamic abnormal scene. If classified as a static normal scene, a low-frequency frame sampling acquisition strategy is executed. Under this strategy, the main control module instructs the image analysis module to enter an intermittent working mode, controlling the image sensor to capture image frames at a preset low sampling rate (e.g., 1 frame / second), simultaneously reducing the image encoding bitrate and scheduling the storage module to a low-power write state. Only lightweight motion detection is performed on the sampled frames to maintain a minimum level of scene perception capability. If classified as a dynamic abnormal scene... In abnormal scenarios, a high-frequency full-frame-rate acquisition strategy is automatically executed. Under this strategy, the main control module instructs the image analysis module to exit intermittent mode, restore the image sensor to full-frame-rate real-time output, simultaneously improve image encoding quality parameters, and trigger the storage module to switch to high-bandwidth write mode to ensure the complete retention of continuous frame data and high-precision input for edge AI real-time analysis in abnormal scenarios. In S203, the edge AI inference model is invoked to perform target detection and semantic segmentation on the acquired dynamic abnormal scene images, identifying crowds, vehicle stagnation, landslide signs, and water level exceeding the line, generating structured analysis data containing the identified object category, confidence level, and coordinate information. In S204, based on the structured analysis... The regional heat distribution parameters of the data, including grid occupancy density, aggregation duration frames, and thermal peak entropy, are weighted and summed to generate a risk index value for the current scene. This risk index value is compared with a preset risk level threshold. If the risk index value is lower than the preset risk level threshold, it is determined to be a general monitoring event. The structured analysis data is only written to the local environmental monitoring log and the system status register is updated, without triggering a remote transmission process. If the risk index value is not lower than the preset risk level threshold, an early warning analysis report is generated. The early warning analysis report includes a timestamp, geographic location tag, and key frame image evidence, and is transmitted back to the rear command center platform through the current preferred communication link.The grid occupancy density value refers to the pixel coverage ratio or instance count density of the identified target (such as people or vehicles) within a unit grid after the monitoring screen is divided into regular grid units, which is used to characterize the degree of spatial aggregation. The aggregation duration frame count refers to the cumulative frame count value of the same grid area or the same target group maintaining a high density state in a continuous video frame sequence, which is used to reflect the duration and stability of the event. The thermal peak entropy value is calculated based on the statistical entropy of the motion intensity or target distribution heat map within the screen, which is used to quantify the disorder of spatial distribution or the intensity of local abrupt changes. The higher the peak entropy value, the more disordered the thermal distribution or the existence of local abnormal aggregation jumps.

[0019] The sensor interface module features standardized interfaces (including RS485, RJ45, and USB interfaces) for connecting various environmental sensors. This enables real-time environmental sensing and early warning, and supports hot-swapping for easy sensor replacement and expansion. Connectable sensors include: temperature sensors (measurement range -40℃~85℃, accuracy ±0.5℃), humidity sensors (measurement range 0~100%RH, accuracy ±3%RH), barometric pressure sensors, vibration sensors (for monitoring earthquakes and landslides), water level sensors (for monitoring floods and waterlogging), and gas sensors (for monitoring toxic and harmful gases). These sensors collect environmental monitoring data and trigger early warnings. The specific steps are as follows: S205 reads real-time monitoring values ​​from the connected sensors through the standardized interface at a preset polling cycle. The read data undergoes validity verification and sliding window mean filtering. Validity verification includes data frame format verification (checking if the read byte sequence conforms to the sensor communication protocol's start bit, data field length, and checksum rules), range overflow verification (determining if the current value exceeds the sensor's nominal measurement range physical limit), and rate of change anomaly verification (for calculation). If the slope of the difference between two consecutive sampled values ​​exceeds the maximum rate of change threshold allowed by the physical characteristics of the sensor, the current sampled value is determined to be an outlier and discarded, and replaced by the previous valid value; S206, the filtered real-time sensor value is compared with a multi-level preset threshold table stored in Flash. The multi-level preset threshold table includes the warning trigger threshold and the alarm cancellation hysteresis value; S207, if any sensor real-time value exceeds the corresponding warning trigger threshold, an environmental warning signal corresponding to the sensor value level is triggered, and the warning event is written to the environmental monitoring log. The environmental monitoring log records the abnormal value, trigger time, and duration; if any sensor real-time value does not reach the alarm cancellation hysteresis value, the normal monitoring state is maintained, no warning event is generated, and the current real-time value is written to the environmental monitoring log as a normal background value record according to the preset storage period; if the current state is under a triggered warning, and the sensor real-time value falls below the alarm cancellation hysteresis value, and the sensor real-time value remains below the alarm cancellation hysteresis value for a preset stable duration, the environmental warning signal is cancelled, a warning cancellation record is generated and written to the environmental monitoring log, and the system returns to the normal monitoring state.

[0020] In this embodiment, the present invention first performs preprocessing such as denoising and illumination compensation on the real-time image stream by adding an image analysis module. Then, it calculates a comprehensive score based on scene dynamic feature parameters such as inter-frame difference amplitude and motion vector distribution entropy value to intelligently distinguish between static normal and dynamic abnormal scenes. It then adaptively switches between low-frequency frame-sampling intermittent acquisition and high-frequency full-frame rate acquisition modes. This reduces the image sampling rate, encoding bitrate, and storage power consumption in normal scenes, while ensuring complete image data retention and high-precision input for edge AI recognition in abnormal scenes. At the same time, it completes target detection and semantic segmentation through an edge AI inference model, calculates a risk index based on regional heat distribution parameters to determine the risk level, and generates an early warning analysis report with timestamps, geographical location, and key frame evidence for high-risk scenes and sends it back to the command center. This achieves accurate judgment and graded early warning of abnormal events on site, and balances monitoring performance, data storage, and equipment power consumption through intelligent acquisition scheduling. This system avoids redundant processing and transmission of invalid data. Furthermore, by adding sensor interface modules with standardized hot-swappable interfaces such as RS485, RJ45, and USB, it can flexibly connect to various environmental sensors, including those for temperature, humidity, vibration, water level, and gas. During data acquisition, triple checks—data frame format, range overflow, and rate of change anomalies—along with sliding window mean filtering, eliminate outliers, ensuring the authenticity and validity of environmental monitoring data. The system then compares and judges the multi-level early warning thresholds and alarm cancellation hysteresis values ​​stored in Flash memory to accurately trigger corresponding environmental warnings and fully record abnormal values, trigger times, and other log information. Simultaneously, the hysteresis value and stability duration mechanism prevents false alarm triggers and frequent jumps. The hot-swappable feature also facilitates rapid sensor replacement and functional expansion. This multi-dimensional environmental perception system comprehensively covers various emergency scenarios such as earthquakes, floods, and toxic gas leaks, significantly improving the comprehensiveness of the system's environmental monitoring, the accuracy of early warnings, and its adaptability to on-site deployment.

[0021] Furthermore, the steps for performing system coordinated control based on feedback signals from each module include: S301 obtains feedback signals from each module in real time through internal bus polling, including voltage fluctuation frequency, signal attenuation slope, protection failure level, and over-temperature event frequency. Specifically: the voltage fluctuation frequency is extracted based on the input voltage / current status word of the power supply module to characterize the stability of power supply quality; the signal attenuation slope is calculated based on the signal strength register value of the communication module to characterize the degradation trend of the communication link; the protection level is evaluated and generated based on the surge protector flag bit of the protection module, and the protection level is divided into normal state, single breakdown state, and permanent failure state according to the set state of the failure flag bit and the cumulative number of set bits; the over-temperature event frequency is generated based on the power amplifier over-temperature flag bit of the audio module to characterize the degree of heat accumulation risk of the power amplifier. S302, based on feedback signals, performs logical synthesis and judgment to generate corresponding collaborative control commands: If the signal attenuation slope parameter of the communication module exceeds the preset attenuation threshold, the intelligent optimization process of the communication link is triggered: the signal quality parameters of each communication unit are obtained and a link scoring matrix is ​​constructed. After eliminating failed links, the channel with the best score is selected to perform transmission takeover, and the power supply of unused radio frequency units is turned off to reduce power consumption; If the voltage fluctuation frequency of the power supply module exceeds the preset fluctuation frequency upper limit, the power supply mode switching process is triggered: based on the mains power detection status and the remaining battery power, the power selection switch is controlled to switch between the mains power path, solar power path, and battery path, and the working mode and power consumption level of each load module are adjusted synchronously; If the protection level of the protection module reaches the protection level threshold, the lightning protection failure alarm process is triggered and a maintenance work order identifier is generated. When the surge protector failure flag is detected and the cumulative number of breakdowns reaches a preset threshold, a maintenance alarm message containing the device identification code and failure timestamp is reported to the command center through the current preferred communication link, and the local maintenance indicator light on the terminal is illuminated; if the frequency of over-temperature events of the audio module exceeds the preset frequency limit, the power amplifier protection derating process is triggered to limit the audio output power to a preset safe power value: a power limiting command is issued to the audio module to forcibly attenuate the maximum output gain of the power amplifier to below the preset safety factor, while an active heat dissipation strategy is activated and the over-temperature flag is continuously monitored. The power limiting is automatically lifted after the temperature drops back to the safe range; S303, a corresponding collaborative control command sequence is generated according to the judgment result, and the command is issued to the corresponding execution module through the bus time-sharing scheduling mechanism to realize the dynamic allocation of system resources and adaptive response to abnormal states.

[0022] In this embodiment, the present invention obtains four types of multi-dimensional feedback signals in real time through internal bus polling: power supply module voltage fluctuation frequency, communication module signal attenuation slope, protection module protection level, and audio module over-temperature event frequency. This accurately characterizes power supply quality stability, communication link degradation trends, equipment protection status, and power amplifier heat accumulation risks, providing precise data for system collaborative control. Based on the above feedback signals, a logical comprehensive judgment is made, triggering differentiated intelligent control processes for different abnormal scenarios: For communication links with signal attenuation slopes exceeding the threshold, a link scoring matrix is ​​constructed, failed channels are eliminated, and the optimal transmission takeover is selected. Simultaneously, the power supply to unused RF units is shut down, ensuring redundancy backup for uninterrupted communication in extreme scenarios while reducing system power consumption. For power supply states with voltage fluctuation frequencies exceeding the upper limit, the system dynamically switches between mains power, solar power, and battery power supply paths based on mains availability and remaining battery power, while simultaneously adjusting the operating mode and power consumption level of each load module. This achieves intelligent optimization of power sources and dynamic adaptation of load power consumption, preventing power instability. The system is designed to prevent system downtime caused by certain factors. For lightning protection failure scenarios where the protection level reaches the threshold, when the cumulative number of lightning arresters breaks down to the preset threshold, a maintenance alarm message containing the device identification code and failure timestamp is reported to the command center via the preferred communication link, and the local maintenance indicator light is illuminated. This enables rapid fault detection and remote early warning, facilitating timely maintenance and repair. For power amplifiers experiencing overheating events exceeding the upper limit, the audio output power is forcibly attenuated to below the safety factor, and an active cooling strategy is activated. Once the temperature returns to the safe range, the power limit is automatically lifted, effectively avoiding the risk of power amplifier damage due to overheating and ensuring stable operation of the audio warning function. Finally, through a bus time-sharing scheduling mechanism, the coordinated control command sequence is precisely sent to the corresponding execution module, achieving dynamic intelligent allocation of system hardware resources, communication resources, and load resources, as well as adaptive response to various abnormal states. This improves the system's reliability, fault tolerance, and environmental adaptability from multiple dimensions, including hardware operation assurance, link stability maintenance, and fault warning and repair, ensuring efficient and stable early warning and scheduling capabilities under various complex emergency conditions.

[0023] Furthermore, the steps for intelligent optimization of communication links based on the communication link status include: S401, acquiring the signal received strength indication value, signal-to-noise ratio value, historical average bit error rate, and link delay jitter value of each communication unit; S402, normalizing the signal received strength indication value and signal-to-noise ratio value, then performing weighted fusion calculation to generate the basic transmission quality index of each communication unit; normalizing the historical average bit error rate and link delay jitter value, then performing weighted fusion calculation to generate the link stability confidence factor of each communication unit; and multiplying and correcting the basic transmission quality index based on the link stability confidence factor to generate the link stability confidence factor of each communication unit. The link health score of the unit; S403, sort the links in descending order according to the link health score, determine the first link of the communication unit in the descending order as the optimal communication link, check whether the current real-time bit error rate of the optimal link is lower than the system set threshold, if so, set the optimal link as the current active transmission channel and shut down the power supply of the other radio frequency units, otherwise issue a communication link abnormality prompt; S404, continuously monitor the link status of the current active transmission channel during data transmission, if a link interruption or bit error rate exceeding the preset bit error rate threshold is detected, perform seamless link switching, select the second best ranked communication link to take over the transmission task, and ensure that data transmission is not lost.

[0024] In this embodiment, the present invention achieves comprehensive and multi-dimensional accurate monitoring of the communication link status by comprehensively collecting four core parameters of each communication unit: signal reception strength indicator (SWIFT), signal-to-noise ratio (SNR), historical average bit error rate (BER), and link delay jitter. SWIFT and SNR characterize the basic transmission quality of the link, while BER and jitter reflect its long-term stability. These four parameters synergistically cover the real-time quality and long-term reliability of the link transmission, providing comprehensive and accurate data support for link optimization. By normalizing the basic transmission parameters (SWIFT and SNR) and stability parameters (BER and jitter), the dimensional differences between different parameters are effectively eliminated, avoiding the influence of single-parameter dimensional bias on the evaluation results. Then, through weighted fusion calculation, a basic transmission quality index and a link stability confidence factor are generated. The stability confidence factor is used to dynamically correct the basic transmission quality index, ultimately obtaining a link health score that accurately reflects the comprehensive performance of each communication unit. This approach addresses the limitations of traditional link optimization methods, which rely on a single parameter and offer only a partial evaluation, ensuring the scientific rigor and rationality of link selection. It selects the optimal link based on a descending order of link health scores, assesses its availability by monitoring the real-time bit error rate (BER) of the optimal link, and designates the qualified optimal link as the active transmission channel while shutting down power to other RF units. This approach fully leverages the transmission advantages of the optimal link while effectively reducing system power consumption, achieving a balance between transmission efficiency and power control. During data transmission, the status of active links is continuously monitored. When a link interruption or excessive BER is detected, the system immediately and seamlessly switches to the next-ranked communication link to take over the transmission task, completely avoiding data loss due to link anomalies. This achieves redundancy and seamless connection of multiple communication units, significantly improving the stability, reliability, and continuity of system communication. It ensures stable transmission of emergency information through intelligent link optimization and seamless switching, regardless of whether ground communication base stations are paralyzed or individual links fail, even in extreme disaster scenarios such as earthquakes and floods, providing timely and reliable communication support for emergency command and decision-making.

[0025] Furthermore, the steps for performing information processing and SM4 encryption based on the received information type include: S501, parsing the header identifier of the received data packet, and classifying the information type into remote control command type, warning text type, firmware upgrade type, and audio stream type based on the data packet protocol type and payload format. Specific steps include: S5011, extracting the protocol identifier field from the transport layer header of the received data packet; if the protocol identifier is CoAP or MQTT-SN lightweight IoT protocol, the payload is initially determined to be a short message of remote control command type or warning text type; if the protocol identifier is HTTP / FTP file transfer protocol, the payload is initially determined to be a firmware upgrade type or audio stream type of large-volume data; S5012, further parsing the application layer payload header. The first byte of the header contains a type code identifier. Type code 0x01 corresponds to the remote control command class, containing system configuration parameter modification and peripheral control command words; type code 0x02 corresponds to the warning text class, containing UTF-8 encoded warning broadcast text and risk level subfields; type code 0x03 corresponds to the firmware upgrade class, containing firmware version number, block index, and firmware data block; type code 0x04 corresponds to the audio stream class, containing audio encoding format identifier and PCM / OPUS audio payload data. S5013, based on the combination of the protocol identifier field and the type code identifier, determines the allocation of received data packets to the corresponding information processing queue and sets differentiated processing priorities for each queue, with the remote control command queue having the highest priority, followed by the audio queue. The stream queue has the second highest priority, followed by the firmware upgrade queue; S502, for remote control command information, the integrated SM4 hardware encryption engine in the main control module is invoked to decrypt the information payload and verify the integrity check code. Specific steps include: S5021, extracting a 16-byte SM4-CBC mode initialization vector and a 32-byte HMAC-SM3 integrity check code from the tail of the received data packet, and loading the initialization vector into the vector register of the SM4 hardware encryption engine; S5022, using a preset device pre-shared key or session negotiation key as the SM4 decryption key, performing CBC mode block decryption on the payload ciphertext data block to obtain the plaintext payload and the PKCS#7 padding identifier field; S5 023. Perform SM3 hash operation on the decrypted plaintext payload to generate a locally calculated checksum. Compare the locally calculated checksum with the HMAC-SM3 integrity checksum carried in the received data packet byte-by-byte at constant time. If they match, the integrity check is deemed to have passed, the security flag is cleared, and the plaintext payload content is extracted to the application buffer. If they do not match, the data packet is deemed to have been tampered with or damaged during transmission. The data packet is discarded, and a security event alarm is reported to the command center. The payload content is extracted after the integrity check is passed. S503. For warning text information, call the national cryptographic hardware encryption engine integrated in the main control module to decrypt the information payload and verify the integrity checksum. The text content is extracted after the integrity check is passed.The text content is input into the speech synthesis engine for natural language processing and prosodic prediction. Through text normalization, syntactic and semantic analysis, and acoustic feature mapping, the engine accurately predicts the pauses, stressed words, and intonation of the speech stream. This ensures that the final synthesized emergency warning broadcast is not only accurate but also conveys a warning tone and sense of urgency that conforms to human auditory habits, significantly improving the alertness and compliance of the crowd. Digital audio waveform data is generated, and a digital filter is used to suppress high-frequency harmonics and reduce noise in the generated audio waveform, producing the audio stream to be broadcast. For firmware upgrade information, after completing SM4 decryption and signature verification of the upgrade package, it is written to the Flash backup partition, and a reboot boot flag is set. A system hot reboot is then performed to complete the firmware upgrade operation. The firmware upgrade operation is an atomic operation process that includes strict security verification, dual-partition backup, and automatic rollback in case of failure. The SM4 decryption and SM3 signature provide dual protection for firmware legitimacy, and the dual-bank alternation mechanism ensures that the system can safely revert to the pre-upgrade state in the event of a power outage or program crash during the upgrade process. This design ensures that emergency warning terminals deployed in remote, uninhabited areas possess extremely high fault tolerance and autonomous recovery capabilities during remote upgrades. For audio stream information, the S505 uses the national cryptographic hardware encryption engine integrated within the main control module to decrypt the audio stream payload and verify its integrity. After the integrity verification passes, the audio stream is decoded, clock synchronized, noise reduced, and gain equalized to generate the audio stream to be played.

[0026] In this embodiment, the present invention accurately classifies four types of information—remote control commands, warning texts, firmware upgrades, and audio streams—by parsing the transport layer protocol identifier and application layer type code of the received data packets. Type code 0x01 corresponds to remote control commands containing system configuration and peripheral control commands; 0x02 corresponds to warning texts containing UTF-8 encoded warning text and risk levels; 0x03 corresponds to firmware upgrades containing firmware versions and data blocks; and 0x04 corresponds to audio streams containing audio encoding and payloads. Simultaneously, differentiated processing queues are allocated to each type of information, and priorities are set (remote control commands are highest, followed by audio streams). (Minimum firmware upgrade), ensuring core commands are executed first, avoiding information processing congestion, and improving information response efficiency in emergency scenarios; through the national cryptographic SM4 hardware encryption engine integrated in the main control module, remote control commands and warning text information are decrypted in groups using SM4-CBC mode, combined with HMAC-SM3 integrity check code byte-by-byte comparison, providing double protection for the confidentiality and integrity of information transmission, effectively preventing data tampering, theft, and transmission damage. If the check is inconsistent, the data packet is discarded and a security alarm is reported, ensuring that only legitimate and valid data enters the processing flow, meeting the security management requirements of emergency information; for For warning text messages, a speech synthesis engine performs natural language processing and prosody prediction to precisely control the pauses, stressed words, and intonation, giving the synthesized speech a warning tone and sense of urgency that conforms to human auditory habits. This significantly improves the alertness and compliance of the crowd. Simultaneously, digital filters suppress high-frequency harmonics and reduce noise, ensuring the warning messages are clear and identifiable. For firmware upgrade messages, after verifying the firmware's legitimacy through SM4 decryption and SM3 signature, the firmware is written to the Flash backup partition and a hot reboot upgrade is performed. A dual-bank alternation mechanism is used to achieve atomic upgrade operations. Even if power outages or program crashes occur during the upgrade process, the system can automatically revert to the state before the upgrade, significantly improving the fault tolerance and autonomous recovery capability of remote upgrades for emergency terminals in remote and uninhabited areas, and preventing system paralysis due to upgrade failures. Overall, through targeted information classification and processing, national-level encryption protection, and differentiated security control, the system not only ensures the security and accuracy of emergency information transmission and processing, but also achieves efficient and orderly processing of various types of information, adapting to the diverse information needs in emergency scenarios. At the same time, it improves the reliability of remote operation and maintenance and upgrades, providing a solid security and efficiency guarantee for the long-term stable operation of the system.

[0027] Furthermore, the steps for triggering corresponding warning actions based on the warning analysis results include: S601, parsing the warning analysis results and extracting the risk level identifier and warning type code carried in the warning analysis results; S602, using the risk level identifier and warning type code as joint index keywords, searching the pre-configured warning response strategy table, and matching the warning level and terminal action set corresponding to the current warning; the warning response strategy table is an association configuration table that maps the relationship between the combination of risk level identifier and warning type code, the warning level, and the terminal action set; S603, if the warning level is the alert level, triggering the audio module to pre-... S604: If the warning level is emergency, the audio module is triggered to output full power to alternate between high-decibel alarm sounds and synthesized speech, while the high-intensity strobe light is controlled to flash at the highest frequency, and all external sensor interfaces in sleep mode are woken up to increase data acquisition density; S605: If the warning level is evacuation level, all terminal actions corresponding to the emergency level are executed, and encrypted data packets are sent to the pre-configured emergency broadcast terminal group through the currently active transmission channel. The encrypted data packets contain the real-time location coordinates of the terminals and evacuation route guidance information.

[0028] In this embodiment, the present invention extracts the risk level identifier and warning type code by parsing the warning analysis results. Using these two as joint index keywords, it retrieves a pre-configured and permanently stored warning response strategy table in Flash memory, and matches and determines the warning level and terminal action set corresponding to the current warning. The warning response strategy table is an associated configuration table that maps the relationship between the combination of risk level identifier and warning type code, the warning level, and the terminal action set. Through a dual-parameter joint index, the present invention achieves precise matching of warning level and action, solving the problem that a single-level index cannot adapt to the differentiated response needs of different warning types at the same level, and completely avoiding the defects of insufficient or excessive warnings, ensuring the pertinence and rationality of the warning response. This invention designs differentiated and refined graded warning actions for different warning levels: For alert-level warnings, the audio module performs periodic voice reminders at a preset low volume, and simultaneously controls the status indicator lights to operate at a preset slow flashing frequency, achieving effective and gentle alerts in low-risk scenarios without causing panic among on-site personnel; For emergency-level warnings, the audio module outputs high-decibel alarm sounds at full power, alternating with synthesized warning voice announcements, and simultaneously controls strong light strobe lights to flash at a preset maximum frequency, achieving strong warnings to on-site personnel, while forcibly waking up all dormant external sensor interfaces to increase the density of environmental data collection, ensuring that on-site personnel can quickly perceive emergency risks and providing high-frequency, real-time environmental data support for emergency command and decision-making; For evacuation-level warnings, in addition to fully executing all emergency-level warning actions, an encrypted data packet containing real-time terminal location coordinates and evacuation route guidance information is simultaneously sent to a pre-configured emergency broadcast terminal group through a preferred communication link, achieving coordinated linkage between warning and evacuation guidance, accurately guiding on-site personnel to evacuate in an orderly manner, effectively avoiding secondary disasters caused by disorderly evacuation, and significantly improving the efficiency and safety of emergency evacuation. This invention, through a tiered early warning system with a full-link design, balances the effectiveness of early warning and the maintenance of on-site order. It links multiple functional modules, including audio, lighting, sensing, and communication, to form a comprehensive and multi-layered early warning response system. This system not only achieves precise adaptation and response to different risk scenarios but also maximizes the synergistic efficiency of multiple modules, significantly improving the emergency response capabilities of the terminal system and providing solid technical support for the safety of on-site personnel and emergency command work.

[0029] Furthermore, the steps for dynamically selecting the power source based on the mains power status and implementing the low-power core protection strategy include: S701, real-time monitoring of the AC mains input status, solar photovoltaic voltage output status, and remaining battery power; S702, if the mains input is detected as valid, the power selection switch is switched to the highest priority mains power supply path, and the solar controller is simultaneously controlled to enter float charging mode, using the solar panels to perform trickle float charging to replenish the battery; S703, if the mains input is detected as invalid, and the solar photovoltaic voltage is higher than the pre-configured photovoltaic start-up voltage threshold, the power selection switch is switched to the independent solar power supply path, and after meeting the current load power consumption requirements of the system, the remaining power is charged into the battery through the charging management chip; S704, if the mains input is detected as invalid, and the solar photovoltaic voltage is higher than the pre-configured photovoltaic start-up voltage threshold, the power selection switch is switched to the independent solar power supply path, and after meeting the current load power consumption requirements of the system, the remaining power is charged into the battery through the charging management chip; If the voltage is lower than the pre-configured photovoltaic start-up voltage threshold and higher than the pre-configured photovoltaic undervoltage cutoff threshold, the control power selection switch is switched to the combined solar and battery power supply path, and the remaining battery power percentage is monitored in real time; S705, if invalid mains input is detected and the solar photovoltaic voltage is lower than or equal to the pre-configured photovoltaic undervoltage cutoff threshold, the control power selection switch is switched to the independent battery power supply path, and the remaining battery power percentage is monitored in real time; S706, when the remaining battery power percentage is detected to be lower than the pre-configured low power warning line, the low power core protection strategy is executed: the high-frequency acquisition function of the image analysis module is turned off, the main frequency of the main control module is reduced, the external auxiliary power output interface is cut off, and only the standby reception function of the L-band satellite communication unit and the emergency alarm broadcast function of the audio module are retained.

[0030] In this embodiment, the present invention achieves dynamic intelligent optimization of power supply sources and full-scenario power supply guarantee by real-time monitoring of AC mains input status, solar photovoltaic voltage output status, and remaining battery power. It adapts to complex outdoor emergency application scenarios without mains power, ensuring both the continuity and stability of system power supply and achieving efficient utilization of clean energy and control of hardware losses. The system is pre-configured with a highest priority switching rule for mains power. When a valid mains input is detected, the power selection switch is switched to the highest priority mains power supply path, and the solar controller is simultaneously controlled to enter float charging mode. Power is drawn from the solar panels to perform trickle float charging to replenish the battery, maintaining the battery at full charge while reducing the need for solar power. The system minimizes power loss in the electrical link, extending the cycle life of both solar panels and batteries. When an invalid mains input is detected and the solar photovoltaic voltage exceeds the pre-configured photovoltaic start-up voltage threshold, the power selection switch is switched to an independent solar power supply path. Priority is given to powering the current system load via solar output. While meeting the system load's power consumption requirements, any remaining energy is charged into the battery through the charging management chip, maximizing the recovery and utilization of clean solar energy and reserving emergency power for extreme scenarios without sunlight or mains power. When an invalid mains input is detected and the solar photovoltaic voltage is below the pre-configured photovoltaic start-up voltage threshold but above the pre-configured photovoltaic undervoltage cutoff threshold, the system automatically switches to a combined solar and battery power supply path. The battery-powered integrated power supply path primarily uses solar power, supplemented by battery charging, to supply power to the system load. This ensures a smooth power transition in low-light scenarios, avoiding system jitter caused by frequent power supply link switching. Simultaneously, it monitors the remaining battery percentage in real time. When invalid AC power input is detected and the solar photovoltaic voltage is below or equal to the pre-configured photovoltaic undervoltage cutoff threshold, it automatically switches to the independent battery power supply path, ensuring uninterrupted power supply in scenarios without AC power or effective solar output. The remaining battery percentage is also monitored in real time. When the remaining battery percentage falls below the pre-configured low-battery warning line, the core low-battery protection strategy is immediately implemented, by disabling the high-voltage protection of the image analysis module. By improving the frequency acquisition function, reducing the operating frequency of the main control module, and cutting off the external auxiliary power supply output interface, the power consumption of non-core loads in the system is significantly reduced. At the same time, only the standby reception function of the L-band satellite communication unit and the emergency alarm broadcast function of the audio module are retained. This maximizes the emergency battery life and ensures that the two core emergency functions of emergency communication standby and emergency warning broadcast are not interrupted in extreme low power scenarios. It completely solves the power supply and battery life dilemma of outdoor emergency terminals when there is no mains power or solar supplementation, avoids the failure of emergency warning function due to power depletion, and ensures that the system can always play a core emergency role in various extreme power supply scenarios. This greatly improves the system's battery life reliability, outdoor environment adaptability and emergency response stability.

[0031] Furthermore, based on the communication link status and signal strength, the steps for performing multi-mode intelligent switching and redundancy backup include: S801, establishing a link status event listener during the system initialization phase to capture the signal strength attenuation slope and link layer retransmission timeout count increment of the communication unit in real time; S802, after normalizing the signal strength attenuation slope and link layer retransmission timeout count increment respectively, performing weighted fusion calculation to generate the link health index of each communication unit, and generating a link health ranking list based on the link health index; S803, when the current working link health index is detected to be lower than the preset switching threshold, the current communication link is recorded as an abnormal link, triggering the switching decision logic, removing the abnormal link from the link health ranking list, and selecting the backup link with the highest link health index and whose corresponding communication standard supports the current transmission data type (text or streaming media); S804, powering on the radio frequency unit of the target backup link and executing the network attach procedure. After successful attachment, the unsent data buffer queue is redirected to the new link to complete the transmission. After the transmission is completed, the radio frequency power supply of the original faulty link is turned off to achieve hot backup redundancy switching. The specific steps of the network attach procedure include: S8041, send an enable signal to the power management chip of the RF unit corresponding to the target backup link, wait for the RF phase-locked loop to stabilize and lock, and read the RF transceiver ready status register to confirm that the RF unit has entered the working state; S8042, according to the communication standard corresponding to the target backup link, execute the network search and cell selection process, perform downlink synchronization on the searched available base station or satellite beam signals, obtain frequency synchronization and time synchronization parameters, and complete the physical layer synchronization establishment; S8043, initiate an attach request or registration request signaling to the network side, carrying the terminal device's unique identifier and authentication. The system responds to parameters, completes bidirectional authentication and key negotiation with the core network or satellite gateway station, and establishes the terminal context; S8044, it initiates a packet data network connection establishment request or service channel allocation request, obtains the IP address or short message address identifier allocated by the network side, completes the bearer link activation, and updates the local routing table to point to the newly established data channel; S8045, after receiving the attach acceptance or registration completion confirmation signaling returned by the network side, it determines that the network attach process has been successfully completed. If no confirmation signaling is received within the preset timeout window, the attach process is re-initiated or the selection of the suboptimal backup link is triggered.

[0032] In this embodiment, the present invention establishes a link status event listener during the system initialization phase to capture the signal strength attenuation slope and link layer retransmission timeout count increment of each communication unit in real time. After normalized weighted fusion calculation, a link health index that accurately reflects the overall link status is generated, and a link health ranking list is generated accordingly. This provides scientific and accurate data support for multi-mode intelligent switching, effectively avoiding switching misjudgments caused by single-parameter evaluation. When the current working link health index is detected to be lower than a preset switching threshold, it is automatically marked as an abnormal link and a switching decision logic is triggered. After removing abnormal links, the backup link with the highest link health and whose communication standard is compatible with the current data type (text or streaming media) is selected first to ensure that the switched link can stably meet the current data transmission requirements and avoid transmission anomalies caused by insufficient link compatibility. A standardized network attachment process (including RF unit activation, physical layer synchronization, bidirectional authentication, and key agreement) is implemented. The system employs several steps, including activating the backup link, to ensure rapid and stable network attachment. Once attachment is successful, the unsent data buffer queue is redirected to the new link for transmission. After transmission is complete, the RF power supply to the original faulty link is shut down, achieving hot backup redundancy switching without data loss or transmission interruption. This reduces system power consumption while ensuring data transmission continuity. Through real-time link status monitoring, precise health assessment, adaptable backup link selection, and seamless hot switching mechanisms, the system enables multi-mode intelligent switching and redundancy backup for each communication unit (4G / 5G, multi-band satellite communication). This completely solves the pain points of traditional emergency communication links being single, easily interrupted, and poorly adaptable. It ensures continuous and stable communication even in complex scenarios such as base station failure, signal attenuation, and extreme disasters, regardless of whether transmitting text or streaming media data. This provides reliable assurance for the real-time transmission of emergency information and significantly improves the system's fault tolerance, continuity, and environmental adaptability.

[0033] Furthermore, the steps for amplifying and broadcasting the received audio signal include: S901, receiving the digital audio data stream sent by the main control module through the bus interface and buffering it in the audio first-in-first-out buffer; S902, converting the digital audio data in the buffer into an analog audio waveform signal through a digital-to-analog converter; S903, inputting the analog audio waveform signal to a preamplifier for voltage amplitude pre-amplification, and filtering out invalid high-frequency noise and power frequency interference outside the human hearing range through a bandpass filter; S904, inputting the filtered signal to a power amplifier for efficient power amplification to drive the voice coil vibration of the waterproof full-range speaker to achieve spatial diffusion of sound waves; S905, during the amplification and broadcasting process, acquiring the background ambient noise decibel value picked up by the feedback microphone through the volume adjustment unit, and based on the background environment... The gain coefficient of the power amplifier is dynamically adjusted based on the noise decibel value to ensure that the voice broadcast sound pressure level is always higher than the preset decibel value of the ambient noise: The target voice sound pressure level is calculated by adding the background ambient noise decibel value to the pre-configured target signal-to-noise ratio threshold. Based on the pre-stored sound pressure level-gain coefficient mapping table, the target voice sound pressure level is converted into the target gain coefficient of the power amplifier, and the target gain coefficient is limited between the pre-configured minimum gain threshold and the maximum gain threshold. The absolute value of the difference between the target gain coefficient and the current working gain coefficient of the power amplifier is judged. Only when the absolute value of the difference exceeds the pre-configured gain adjustment hysteresis window is the target gain coefficient sent to the power amplifier to complete the gain update. After the update, the real-time sound pressure level of the speaker output is checked to ensure that the voice broadcast sound pressure level is always higher than the pre-configured decibel difference of the ambient noise.

[0034] In this embodiment, the present invention receives the digital audio data stream sent by the main control module through the bus interface and buffers it in the audio first-in-first-out buffer to ensure the continuity of audio data transmission and avoid broadcast stuttering. A digital-to-analog converter accurately converts the digital audio data into an analog audio waveform signal, providing a high-quality base signal for subsequent amplification. After pre-amplification by a preamplifier, the analog audio signal is filtered by a bandpass filter to remove invalid high-frequency noise and power frequency interference outside the human hearing range, effectively improving the clarity of the audio broadcast and preventing noise interference from affecting the on-site personnel's recognition of the warning information. The filtered signal is then efficiently amplified by a power amplifier to drive a waterproof full-range speaker to achieve sound wave spatial diffusion. Its waterproof characteristics make it suitable for complex emergency scenarios such as outdoor rain, snow, and humidity. This system ensures stable speaker operation. During broadcasting, the volume control unit uses a feedback microphone to pick up the background ambient noise level in decibels and dynamically adjusts the gain coefficient of the power amplifier. This ensures that the sound pressure level of the voice broadcast is always higher than the preset ambient noise level, completely solving the problem of unclear and inaudible broadcasts caused by noisy emergency environments. The system employs a complete audio signal processing, noise reduction filtering, power amplification, and adaptive volume control mechanism to ensure clear, continuous, and stable warning audio broadcasts. It is adaptable to outdoor emergency scenarios with varying noise levels, allowing all personnel on-site to accurately and promptly receive warning information. This maximizes the warning effect of audio alerts, buys valuable time for emergency response, and enhances the practicality and reliability of the system's warning broadcasts.

[0035] like Figure 2 The flowchart shown is a flowchart of an intelligent emergency early warning method based on multi-network fusion and edge computing provided in this application embodiment. It includes the following steps: After the system is powered on and initialized, firmware configuration is loaded, full module self-test is performed, multi-mode communication link is attached, and storage module is initialized, it enters normal monitoring mode. Under normal monitoring, the main control module performs low-power scheduling, the image analysis module performs low-frequency frame sampling and lightweight motion detection, the sensor interface module periodically collects environmental data and completes validity verification and sliding window filtering, and simultaneously monitors the communication link status and power supply status. When the data exceeds the threshold, an early warning response is initiated. During the early warning response, high-frequency full-frame rate image acquisition is switched, an early warning analysis report with timestamp and location tag is generated, abnormal sensor data is integrated, encrypted with national cryptographic standards, and transmitted back through the optimal communication link. High-bandwidth data storage is initiated simultaneously. The early warning information is parsed and a graded early warning strategy is matched. The corresponding power amplification alarm and adaptive volume adjustment are executed. The power supply source is dynamically monitored, and the low-power core protection strategy is executed when the battery is low. Non-core power consumption is reduced to ensure the core functions of communication and alarm. After the early warning is lifted, normal low-power monitoring is restored, the link health is monitored in real time, and multi-mode communication seamless switching and redundant backup takeover are executed when the link fails.

Claims

1. An intelligent emergency early warning system based on multi-network fusion and edge computing, characterized in that, It includes a power supply module, a protection module, a communication module, and an audio module. These modules are connected via a bus and are controlled and scheduled uniformly by a main control module. The main control module performs system collaborative control based on feedback signals from each module. The system collaborative control includes power supply mode switching, image analysis, sensor data acquisition, intelligent optimization of communication links based on communication link status, and information processing and national cryptographic encryption based on the type of received information. The information processing includes speech synthesis, noise reduction filtering, and early warning analysis. The corresponding early warning action is triggered based on the early warning analysis result. The power supply module includes a solar power supply unit, a battery power supply unit, and a mains power charging unit, which are used to dynamically select the power source based on the mains power status and execute the core low power protection strategy. The protection module includes shell protection, lightning protection and impact protection, which are used to achieve integrated sealed protection of the terminal; The communication module includes a 4G / 5G communication unit, an L-band satellite communication unit, an S-band satellite communication unit, and a Ku-band satellite communication unit, which are used to perform multi-mode intelligent switching and redundancy backup based on the communication link status and signal strength. The audio module includes a digital-to-analog converter, a preamplifier, a power amplifier, a speaker, and a volume control unit, used to amplify and broadcast the received audio signal.

2. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 1, characterized in that, It also includes an image analysis module and a sensor interface module. The image analysis module is used to perform hierarchical intelligent data acquisition, edge AI recognition and analysis based on scene status and on-site image content, and generate an early warning analysis report to be sent back to the command center. The specific steps are as follows: S201, The image sensor configured in the image analysis module continuously captures the real-time image stream of the monitored area, and performs preprocessing operations on the continuous frame images. The preprocessing operations include image denoising, illumination compensation and size normalization. S202, based on the scene dynamic feature parameters of the preprocessed image, perform weighted summation processing to obtain the scene dynamic comprehensive score value, and dynamically determine the scene type based on the scene dynamic comprehensive score value. The scene dynamic feature parameters include inter-frame difference magnitude, motion vector distribution entropy value and regional pixel change rate. If the scene is determined to be a static, routine scene, a low-frequency frame sampling strategy will be implemented. If the scene is determined to be a dynamic abnormal scene, the high-frequency full frame rate acquisition strategy will be automatically executed. S203 performs target detection and semantic segmentation recognition on the collected dynamic abnormal scene images to generate structured analysis data; S204. Based on the regional heat distribution parameters of the structured analysis data, including grid occupancy density value, aggregation duration frame count, and thermal peak entropy value, a weighted summation process is performed to generate a risk index value for the current scene. The risk index value is compared with a preset risk level threshold. If the risk index value is lower than the preset risk level threshold, the remote feedback process is not triggered. If the risk index value is not lower than the preset risk level threshold, an early warning analysis report is generated. The sensor interface module is used to collect environmental monitoring data and trigger early warnings. The specific steps are as follows: S205, read the real-time monitoring values ​​of the access sensor, and perform validity verification and sliding window mean filtering on the read data. The validity verification includes data frame format verification, range overflow verification, and rate of change abnormality verification. S206, compare the filtered real-time sensor values ​​with a multi-level preset threshold table, which includes a warning trigger threshold and an alarm cancellation hysteresis value. S207, if the real-time value of any sensor exceeds the corresponding warning trigger threshold, an environmental warning signal corresponding to the level of the sensor value is triggered, and the warning event is written into the environmental monitoring log, which records the abnormal value, the trigger time, and the duration. If the real-time value of any sensor does not reach the alarm cancellation hysteresis value, the normal monitoring state will be maintained and no early warning event will be generated. If the current state is under an alert, and the real-time sensor value falls below the alarm cancellation hysteresis value, and remains below the alarm cancellation hysteresis value for a preset stable period of time, then the environmental alert signal will be cancelled, an alert cancellation record will be generated and written to the environmental monitoring log, and the system will return to normal monitoring status.

3. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 1, characterized in that, The steps for performing system coordinated control based on feedback signals from each module include: S301 obtains feedback signals from each module in real time through internal bus polling, including voltage fluctuation frequency, signal attenuation slope, protection failure level, and over-temperature event frequency; S302, based on the feedback signal, perform logical synthesis and judgment to generate corresponding collaborative control instructions: if the signal attenuation slope of the communication module exceeds the preset attenuation threshold, trigger the intelligent optimization process of the communication link; if the voltage fluctuation frequency of the power supply module exceeds the preset fluctuation frequency upper limit, trigger the power supply mode switching process; if the protection level of the protection module reaches the preset protection level threshold, trigger the lightning protection failure alarm process and generate a maintenance work order identifier; if the frequency of over-temperature events of the audio module exceeds the preset frequency upper limit, trigger the power amplifier protection derating process to limit the audio output power to the preset safe power value. S303 generates a corresponding sequence of collaborative control instructions based on the judgment result, and sends the instructions to the corresponding execution module through a bus time-sharing scheduling mechanism to achieve dynamic allocation of system resources and adaptive response to abnormal states.

4. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 1, characterized in that, The steps for performing intelligent optimization of communication links based on communication link status include: S401, respectively acquire the signal reception strength indication value, signal-to-noise ratio value, historical average bit error rate and link delay jitter value of each communication unit; S402, after normalizing the signal received strength indication value and the signal-to-noise ratio value, a weighted fusion calculation is performed to generate the basic transmission quality index of each communication unit. After normalizing the historical average bit error rate and the link delay jitter value, a weighted fusion calculation is performed to generate the link stability confidence factor of each communication unit. Based on the link stability confidence factor, the basic transmission quality index is dynamically corrected to generate the link health score of each communication unit. S403: Based on the link health score, perform descending sorting, determine the first communication unit link in the descending sort as the optimal communication link, check whether the current real-time bit error rate of the optimal communication link is lower than the system set threshold, if so, set the optimal communication link as the current active transmission channel and turn off the power supply of the other radio frequency units, otherwise issue a communication link abnormality prompt. S404 continuously monitors the status of the currently active transmission channel link during data transmission. If a link interruption or bit error rate exceeding the preset bit error rate threshold is detected, seamless link switching is performed, and the next best-ranked communication link is selected to take over the transmission task, ensuring that data transmission is not lost.

5. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 1, characterized in that, The steps for performing information processing and national cryptographic encryption based on the type of received information include: S501 parses the header identifier of the received data packet and classifies the information type into remote control command type, warning text type, firmware upgrade type and audio stream type based on the data packet protocol type and payload format; S502, for remote control command information, calls the national cryptographic hardware encryption engine integrated in the main control module to decrypt the information payload and verify the integrity check code. After the integrity check is passed, the effective payload content is extracted. S503, for warning text information, calls the national cryptographic hardware encryption engine integrated in the main control module to decrypt the information payload and verify the integrity check code. After the integrity check is passed, the text content is extracted. The extracted text content is input into the speech synthesis engine, and natural language processing and prosody prediction are performed in sequence to generate digital audio waveform data. The digital filter is called to perform high frequency harmonic suppression and noise reduction on the generated audio waveform to generate the audio stream to be broadcast. For firmware upgrade information, S504 calls the national cryptographic hardware encryption engine integrated in the main control module to decrypt and verify the signature of the upgrade package. After successful verification, the upgrade package is written to the system backup partition, and a system hot reboot is performed to complete the firmware upgrade operation. For audio stream information, the S505 calls the national cryptographic hardware encryption engine integrated in the main control module to decrypt the audio stream payload and verify the integrity check code. After the integrity check is passed, the audio stream is decoded, clock synchronization is calibrated, noise is reduced and gain is equalized to generate the audio stream to be played.

6. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 5, characterized in that, The steps for triggering the corresponding early warning action based on the early warning analysis result include: S601, parse the early warning analysis result and extract the risk level identifier and early warning type code carried in the early warning analysis result; S602, using the risk level identifier and the warning type code as joint index keywords, retrieve the pre-configured warning response strategy table, and match the warning level and terminal action set corresponding to the current warning; the warning response strategy table is an association configuration table that maps the correspondence between the combination of risk level identifier and warning type code, warning level, and terminal action set; S603, if the warning level is the prompt level, trigger the audio module to perform periodic voice reminders at a preset low volume, and at the same time control the status indicator light to work at a preset slow flashing frequency; S604, if the warning level is emergency, triggers the audio module to output at full power to alternate between high-decibel alarm sound and synthesized voice, while controlling the strong light strobe to flash at the highest frequency, and wakes up all external sensor interfaces in sleep mode to increase data acquisition density. S605, if the warning level is evacuation level, execute all terminal actions corresponding to the emergency level, and at the same time send encrypted data packets to the pre-configured emergency broadcast terminal group through the currently active transmission channel. The encrypted data packets contain the real-time location coordinates of the terminals and evacuation route guidance information.

7. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 1, characterized in that, The steps of dynamically selecting the power source based on the mains power status and implementing the core low-power protection strategy include: S701 monitors the AC input status of mains power, the output voltage status of solar photovoltaic power, and the remaining battery power in real time. S702 If a valid AC power input is detected, the power selection switch is switched to the highest priority AC power supply path, and the solar controller is switched to float charging mode to perform trickle float charging to replenish the battery by drawing power from the solar panel. S703 If an invalid mains input is detected and the solar photovoltaic voltage is higher than the pre-configured photovoltaic start-up voltage threshold, the power selection switch is controlled to switch to the independent solar power supply path. After meeting the current load power consumption requirements of the system, the remaining power is charged into the battery through the charging management chip. S704 If an invalid mains power input is detected, and the solar photovoltaic voltage is lower than the pre-configured photovoltaic start-up voltage threshold and higher than the pre-configured photovoltaic undervoltage cutoff threshold, the power selection switch is switched to a combined solar and battery power supply path, and the remaining battery charge percentage is monitored in real time. S705 If an invalid mains input is detected and the solar photovoltaic voltage is lower than or equal to the pre-configured photovoltaic undervoltage cutoff threshold, the power selection switch is switched to the independent power supply path of the battery, and the remaining battery power percentage is monitored in real time. When the S706 detects that the remaining battery power percentage is lower than the pre-configured low power warning line, it executes the core low power protection strategy: disables the high-frequency acquisition function of the image analysis module, reduces the main control module's operating frequency, cuts off the external auxiliary power supply output interface, and retains only the standby reception function of the L-band satellite communication unit and the emergency alarm broadcast function of the audio module.

8. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 1, characterized in that, The steps for performing multi-mode intelligent switching and redundancy backup based on communication link status and signal strength include: S801 establishes a link status event listener during the system initialization phase to capture the signal strength attenuation slope of the communication unit and the link layer retransmission timeout count increment in real time. S802, after normalizing the signal strength attenuation slope and the link layer retransmission timeout count increment respectively, a weighted fusion calculation is performed to generate the link health index of each communication unit, and a link health ranking list is generated based on the link health index. S803: When the current working link health index is detected to be lower than the preset switching threshold, the current communication link is recorded as an abnormal link, the switching decision logic is triggered, the abnormal link is removed from the link health sorting list, and the backup link with the highest link health index and whose corresponding communication mode supports the current data type is selected. S804 powers on the RF unit of the target backup link and executes the network attach procedure. After successful attachment, the untransmitted data buffer queue is redirected to the new link to complete the transmission. After the transmission is completed, the RF power supply of the original faulty link is turned off, realizing hot backup redundancy switching.

9. The intelligent emergency early warning system based on multi-network fusion and edge computing as described in claim 1, characterized in that, The step of amplifying and broadcasting the received audio signal includes: S901 receives digital audio data streams from the main control module via the bus interface and buffers them in the audio first-in-first-out buffer. S902 converts digital audio data in the buffer into analog audio waveform signals using a digital-to-analog converter. S903 inputs the analog audio waveform signal to the preamplifier for voltage amplitude pre-amplification, and filters out invalid high-frequency noise and power frequency interference through the bandpass filter; S904 inputs the filtered signal to the power amplifier for efficient power amplification, thereby driving the voice coil of the waterproof full-range loudspeaker to vibrate and achieve spatial diffusion of sound waves; S905, during the amplified broadcast process, the background ambient noise decibel value picked up by the feedback microphone is collected through the volume adjustment unit, and the gain coefficient of the power amplifier is dynamically adjusted based on the background ambient noise decibel value to ensure that the sound pressure level of the voice broadcast is always higher than the preset ambient noise decibel value.

10. An intelligent emergency early warning method based on multi-network fusion and edge computing, applied to the intelligent emergency early warning system based on multi-network fusion and edge computing as described in any one of claims 1-9, characterized in that, Includes the following steps: After the system powers on and completes initialization, firmware configuration loading, full module self-test, multi-mode communication link attachment, and storage module initialization, it enters normal monitoring mode. Under normal monitoring, the main control module performs low-power scheduling, the image analysis module performs low-frequency frame sampling and lightweight motion detection, the sensor interface module periodically collects environmental data and completes validity verification and sliding window filtering, and synchronously monitors the communication link status and power supply status. When the data exceeds the threshold, an early warning response is initiated. When responding to an early warning, switch to high-frequency full-frame-rate image acquisition, generate an early warning analysis report with timestamps and location tags, integrate abnormal sensor data, encrypt it with national cryptographic standards, and transmit it back through the optimal communication link, while simultaneously starting high-bandwidth data storage. Analyze the early warning information and match the graded early warning strategy, execute the corresponding power amplification alarm and adaptive volume adjustment, dynamically monitor the power supply source and execute the low power core protection strategy when the battery is low, and reduce non-core power consumption to ensure communication and alarm core functions. After the warning is lifted, normal low-power monitoring is restored, link health is monitored in real time, and seamless switching of multi-mode communication and redundant backup takeover are performed when the link fails.

Images