Multi-channel earthquake early warning emergency linkage system of Internet of Things

Abstract

The invention discloses an Internet of Things multichannel earthquake early warning emergency linkage system, belongs to the technical field of earthquake early warning, and aims to solve the problems of insufficient real-time performance, reliability, scene coverage and safety in the prior art. The sensing layer collects data cooperatively through various sensors, eliminates environmental interference and ensures high-precision seismic wave detection, the edge calculation layer filters noise in real time by adopting an advanced algorithm, and confirms seismic events through multi-dimensional comparison, so that the local processing efficiency is remarkably improved, the cloud load is greatly reduced, a cloud platform fuses multi-source data, and the seismic detection efficiency is improved. A machine learning model is combined to dynamically optimize an early warning threshold value, the accuracy of earthquake type classification and intensity prediction is improved, full-link optimization from data acquisition to intelligent decision making is realized, a multi-channel distribution module is designed, instructions are transmitted according to priority levels, and urban and rural full-scene coverage is ensured through multiple communication technologies.

Description

Technical Field

[0001] The present invention relates to the technical field of earthquake early warning, and particularly to an Internet of Things multi-channel earthquake early warning emergency linkage system. Background Art

[0002] Earthquake early warning is a disaster reduction technology that sends an alarm to potential disaster-stricken areas after an earthquake occurs and before destructive seismic waves arrive, by taking advantage of the difference in the propagation speed of seismic waves and the fast transmission characteristics of electromagnetic waves. Through the strategy of "exchanging space for time", earthquake early warning cannot completely eliminate the earthquake risk, but significantly reduces the disaster losses. Its effectiveness depends on technology optimization, infrastructure improvement, and the enhancement of public emergency awareness.

[0003] Published patent: An earthquake early warning information broadcast system (Publication No.: CN119049228A), belonging to the field of earthquake early warning, solves the problems of how to achieve early detection of seismic waves during an earthquake to extend the early warning response time, and at the same time make a safer and more effective strategy for the emergency stop of elevators when only the arrival of P waves is recognized; the system includes: an earthquake monitoring and analysis module analyzes and judges whether the operation of the elevators in the target building complex will be threatened by the earthquake according to the seismic parameter time series data detected by several earthquake monitoring devices arranged around the target building complex, and generates an early warning signal of the corresponding seismic wave type; the elevator earthquake early warning analysis module obtains the early warning signal, analyzes and obtains the operation control instructions of the corresponding elevators in the target building complex, and sends the obtained operation control instructions of the corresponding elevators to the elevator controllers of the corresponding elevators, and at the same time generates an earthquake early warning prompt signal and sends it to the voice broadcast device of the corresponding elevator.

[0004] It only focuses on the emergency control of elevators, does not involve the linkage response of urban lifeline systems such as subways and power, has a single scenario coverage, cannot achieve multi-facility collaborative disaster prevention, lacks multi-sensor collaborative calibration, is vulnerable to environmental interference, the collected data has great limitations, needs to directly upload the original data to the cloud for processing, has a high delay, is difficult to meet the second-level early warning requirements, only relies on a single network channel to distribute early warnings, cannot cover areas without network, does not pre-store a local emergency instruction library, and cannot trigger key operations such as elevator stops when the network is disconnected, and the system reliability is insufficient. Summary of the Invention

[0005] The purpose of the present invention is to provide an Internet of Things multi-channel earthquake early warning emergency linkage system, which solves the problems of the deficiencies in real-time performance, reliability, scenario coverage, and security of the prior art.

[0006] To achieve the above object, the present invention provides the following technical solutions: An Internet of Things multi-channel earthquake early warning emergency linkage system, including an Internet of Things perception layer, an edge computing layer, a cloud platform analysis module, a multi-channel distribution module, a dynamic fault tolerance module, an emergency linkage module, a self-checking and fault recovery module, and a user interaction interface. The Internet of Things perception layer is responsible for the acquisition and preliminary processing of seismic wave data. The edge computing layer performs noise filtering, waveform comparison, and strategy pre-computation locally to reduce the cloud load. The cloud platform analysis module realizes global optimization through machine learning and dynamic threshold adjustment and generates encryption parameters. The multi-channel distribution module ensures that the early warning information reaches different scenarios through multi-path communication. The dynamic fault tolerance module monitors the network status and performs redundancy checks to ensure the reliability of the system under abnormal conditions. The emergency linkage module controls elevators, subways, and power equipment according to the early warning instructions to achieve a fast and safe response. The self-checking and fault recovery module includes a health monitoring unit, a fault switching and recovery unit, and a log and feedback unit. The self-checking and fault recovery module automatically detects the health status of the sensors every 24 hours and switches to the standby sensors in case of a fault. The user interaction interface provides a visual early warning map and emergency guidance, supporting functions such as voice broadcast, vibration reminder, and multi-language switching.

[0007] Further, the Internet of Things perception layer includes low-power seismic monitoring sensors and a data acquisition unit. The low-power seismic monitoring sensors include a three-axis accelerometer, a geomagnetic sensor, and a temperature compensation unit. The data acquisition unit includes a Zigbee module and an analog-to-digital converter. The three-axis accelerometer, the geomagnetic sensor, and the temperature compensation unit are all deployed on building foundations, bridge nodes, and subway tunnels. The ground vibration and geomagnetic field changes are detected in real time through the three-axis accelerometer and the geomagnetic sensor. The three-axis accelerometer captures three-dimensional vibration signals based on MEMS technology. The geomagnetic sensor monitors geomagnetic anomalies. The temperature compensation unit automatically calibrates the influence of the ambient temperature on the sensor data to ensure the measurement accuracy. The data acquisition unit converts the analog signal output by the sensor into a digital signal through the analog-to-digital converter. The Zigbee module transmits the data to the edge computing layer through a low-power wireless protocol and supports self-organizing networks to expand the coverage range.

[0008] Furthermore, the edge computing layer includes a noise filtering module, a waveform comparison module, and a local cache unit. The noise filtering module includes an input layer, a convolutional neural network, and an output layer. The waveform comparison module includes an amplitude comparison unit, a frequency comparison unit, and a period similarity unit. The local cache unit includes an FPGA chip and a pre-computation strategy library. The noise filtering module uses a convolutional neural network to remove noise from the original seismic waveform. The input layer receives data, the convolutional layer extracts features, and the output layer marks valid signals. The waveform comparison module confirms the type of seismic wave through three comparisons. The amplitude comparison unit calculates the amplitude difference between adjacent sensors, the frequency comparison unit analyzes the main frequency difference, and the period similarity unit calculates the waveform cosine similarity. The FPGA chip stores the pre-computed elevator docking strategy and the prediction of the seismic wave arrival time. The pre-computation strategy library caches the elevator docking plan and directly calls the local strategy to execute emergency operations when the network is disconnected.

[0009] Furthermore, the cloud platform analysis module includes a machine learning model, a dynamic threshold adjustment unit, and an encryption parameter generation unit. The machine learning model includes a support vector machine and a random forest model. The dynamic threshold adjustment unit includes a geological data interface, an amplitude threshold generator, and a frequency threshold generator. The encryption parameter generation unit includes an SM3 module and an SM4 key generator. The support vector machine and the random forest model fuse multi-source data, classify the seismic type, and predict the affected range. The support vector machine classifies through a hyperplane, and the random forest votes through multiple decision trees. The warning thresholds of amplitude and frequency are dynamically adjusted according to geological data to adapt to different regional geological conditions. The national secret SM3 hash algorithm is used to compress data, and the SM4 dynamically generates a 256-bit symmetric encryption key. The key synchronization is achieved through the blockchain smart contract to prevent man-in-the-middle attacks.

[0010] Furthermore, the multi-channel distribution module includes a communication unit and an offline device wake-up unit. The communication unit includes a Beidou transceiver, a 5G interface, and a digital broadcast transmitter. The offline device wake-up unit includes a Beidou instruction encoder and a rural emergency broadcast terminal. The communication unit covers the network-free area through Beidou satellite short messages. The 5G interface preferentially pushes warnings to key facilities. The digital broadcast transmitter issues evacuation guidelines to the public. The offline device wake-up unit wakes up the rural emergency broadcast terminal by sending specific frequency band instructions through the Beidou satellite, and the terminal broadcasts the pre-stored voice warning in the offline state.

[0011] Further, the dynamic fault tolerance module includes a network status monitoring unit, a redundancy check unit, and a local emergency instruction library. The network status monitoring unit includes a packet loss rate detector and a delay detector. The redundancy check unit includes a CRC32 checker and a data retransmission controller. The local emergency instruction library includes an elevator docking plan library and a subway braking instruction library. The network status monitoring unit real-time detects the packet loss rate and delay of the 4G / 5G network. After the packet loss rate and delay reach the set values, it triggers the communication channel switching. The redundancy check unit verifies the data integrity through the CRC32 check algorithm and automatically retransmits when the check fails to ensure the transmission reliability. The local emergency instruction library stores pre-computed elevator docking plans and subway braking instructions, which are directly called and executed when the network is disconnected to avoid relying on the real-time network.

[0012] Further, the emergency linkage module includes an elevator intelligent docking unit, a subway braking unit, and a power cut-off unit. The elevator intelligent docking unit includes a laser rangefinder and an elevator controller. The subway braking unit includes a braking signal generator and a hydraulic braking actuator. The power cut-off unit includes an intelligent circuit breaker and a power status feedback module. The elevator intelligent docking unit real-time locates the elevator position through the laser rangefinder, and the elevator controller moves the elevator to the nearest safe floor within 2 seconds. The subway braking unit triggers the hydraulic braking system through the braking signal generator and completes the emergency braking within 3 seconds. After receiving the instruction through the intelligent circuit breaker, the power cut-off unit cuts off the power supply in the high-risk area within 1 second, and the power status feedback module real-time transmits back the power grid status.

[0013] Further, the system operation process is as follows: S1. Data collection and preliminary processing S1.1 Sensor network startup: The low-power seismic monitoring sensors deployed on the building foundation, bridge, and subway tunnel real-time collect ground vibration, geomagnetic field change, and environmental temperature data; The sensor eliminates environmental interference through the temperature compensation unit to ensure data accuracy; S1.2 Signal conversion and transmission: The data collection unit converts the analog signal output by the sensor into a digital signal through an analog-to-digital converter, and the digital signal is transmitted to the edge computing layer through the Zigbee module using a low-power wireless protocol, supporting self-organizing network to expand the coverage range; S2. Edge computing layer real-time processing S2.1 Noise filtering: The noise filtering module in the edge computing layer uses a convolutional neural network to eliminate noises such as traffic vibration and mechanical interference, and retains the effective seismic signals; S2.2 Waveform comparison and confirmation: The waveform comparison module performs three comparisons: Amplitude comparison: Calculate the amplitude difference between adjacent sensors to exclude local interference; Frequency comparison: Extract the main frequency through the fast Fourier transform to determine whether it is a seismic event; Period similarity comparison: Calculate the waveform cosine similarity to confirm the consistency of multi-sensor data; If all three comparisons pass, it is determined as a valid seismic wave; S2.3 Local policy pre-storage: The local cache unit stores the pre-computed emergency policies through the FPGA chip and directly calls them when the network is disconnected; S2.4 Data upload: The processed key data is uploaded to the cloud platform analysis module; S3. Cloud platform global optimization and instruction generation S3.1 Multi-source data fusion: The machine learning model fuses sensor data, geological database, and historical earthquake data to classify earthquake types and predict intensity and impact range; S3.2 Dynamic threshold adjustment: Dynamically optimize the amplitude and frequency warning thresholds according to geological conditions to reduce the false alarm rate; S3.3 Encryption and instruction generation: The encryption parameter generation unit encrypts data using the national secret SM3 / SM4 algorithm and synchronizes the keys to the edge nodes in combination with blockchain technology; Generate global warning instructions; S4. Multi-channel redundant distribution S4.1 Priority distribution strategy: The communication unit pushes instructions according to priorities: First-level priority: Real-time transmission through the 5G private network; Second-level priority: Release evacuation guidelines through digital broadcasting; Third-level priority: Send encrypted short messages through Beidou satellites to wake up the emergency broadcast terminal; S4.2 Wake-up of offline devices: The Beidou instruction encoder sends specific encoded instructions to wake up the rural emergency broadcast terminal and broadcast the pre-stored voice warnings; S5. Dynamic fault tolerance and emergency execution S5.1 Network anomaly handling: The dynamic fault tolerance module monitors the network packet loss rate and latency in real time: When the main network is abnormal, automatically switch to the Beidou satellite or local broadcast link; The redundancy check unit ensures the integrity of the instructions through CRC32 check and automatic retransmission; S5.2 Emergency response to network disconnection: The local emergency instruction library calls the pre-stored policies to directly trigger the operations of elevators, subways, and power equipment; S5.3 Terminal device linkage: The emergency linkage module performs operations after receiving instructions: Intelligent elevator docking: The laser rangefinder locates the elevator position and docks it to a safe floor; Subway hierarchical braking: Generate hierarchical braking signals to complete emergency braking; Power cut-off: Cut off the power supply in high-risk areas and retain the emergency power supply for hospitals and subways; S6. User interaction and feedback closed-loop S6.1 Multi-modal warning reach: The user interface publishes warning information to the public through voice announcements, vibration reminders, and graphic and text push, and supports one-key navigation to shelters; S6.2 Visual map display: Dynamically render the epicenter location, evacuation routes, and safe areas, and adapt to large-screen command centers and mobile terminals; S6.3 User feedback collection: Users report their status through interface buttons or voice commands, and the data is encrypted and transmitted back to the cloud platform to optimize strategies; S7. System health management and maintenance S7.1 Periodic self-check: The self-check and fault recovery module detects the status of sensors, networks, and hardware, and warns of low battery or overload; S7.2 Automatic fault switching: When the sensor fails, it switches to the standby node, and when the process crashes, it automatically restarts the service; S7.3 Logging and optimization: Encrypt and store fault logs, and push maintenance suggestions in combination with machine learning models.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The present invention constructs a hierarchical architecture with the linkage of the Internet of Things perception layer, edge computing layer, and cloud platform analysis module. The perception layer cooperatively collects data through multiple sensors, eliminates environmental interference, and ensures high-precision seismic wave detection. The edge computing layer uses advanced algorithms to filter noise in real time, and confirms seismic events through multi-dimensional comparison. The local processing efficiency is significantly improved, and the cloud load is greatly reduced. The cloud platform integrates multi-source data, combines machine learning models to dynamically optimize the warning threshold, improves the accuracy of earthquake type classification and intensity prediction, and realizes the full-link optimization from data collection to intelligent decision-making.

[0015] 2. The system of the present invention designs a multi-channel distribution module, transmits instructions hierarchically according to priorities, ensures full-scenario coverage in urban and rural areas through multiple communication technologies, the dynamic fault tolerance module monitors the network status in real time, automatically switches to the standby link when the main network is abnormal, and ensures the integrity of instructions through the verification and retransmission mechanism. The local emergency instruction library pre-stores key strategies and directly calls and executes operations when the network is disconnected. Combined with encryption technology to synchronize keys, seamless emergency response under network interruption is realized.

[0016] 3. The emergency linkage module of the present invention realizes rapid response through precise terminal control, ensures the safe docking of elevators, hierarchical braking of subways, and power cut-off in high-risk areas. The user interface integrates dynamic map rendering, supports multi-modal warnings and collects user feedback to optimize strategies. The self-check and fault recovery module periodically detects the system status, automatically switches to the standby node or restarts the service, forming a full-closed-loop earthquake-resistant system, significantly improving the real-time performance, reliability of the urban lifeline system and the public's emergency response ability. Description of the drawings

[0017] Figure 1 This is the system architecture diagram of the present invention; Figure 2 This is the system flowchart of the present invention. Detailed implementation manners

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] To solve the problems of lack of real-time performance, reliability, scene coverage, and security, as Figure 1-2 shown, the following preferred technical solutions are provided: The Internet of Things multi-channel earthquake early warning emergency linkage system includes an Internet of Things perception layer, an edge computing layer, a cloud platform analysis module, a multi-channel distribution module, a dynamic fault tolerance module, an emergency linkage module, a self-checking and fault recovery module, and a user interaction interface. The Internet of Things perception layer is responsible for the acquisition and preliminary processing of seismic wave data. The edge computing layer performs noise filtering, waveform comparison, and strategy pre-computation locally to reduce the load on the cloud. The cloud platform analysis module realizes global optimization through machine learning and dynamic threshold adjustment and generates encryption parameters. The multi-channel distribution module ensures that the early warning information reaches different scenarios through multi-path communication. The dynamic fault tolerance module monitors the network status and performs redundant checks to ensure the reliability of the system under abnormal conditions. The emergency linkage module controls elevators, subways, and power equipment according to early warning instructions to achieve fast and safe responses. The self-checking and fault recovery module includes a health monitoring unit, a fault switching and recovery unit, and a log and feedback unit. The self-checking and fault recovery module automatically detects the health status of sensors every 24 hours and switches to standby sensors in case of faults. The user interaction interface provides a visual early warning map and emergency guidance, supporting functions such as voice broadcast, vibration reminder, and multi-language switching.

[0020] The Internet of Things perception layer includes low-power seismic monitoring sensors and data acquisition units. The low-power seismic monitoring sensors include triaxial accelerometers, geomagnetic sensors, and temperature compensation units. The data acquisition units include Zigbee modules and analog-to-digital converters. As the data acquisition front-end of the earthquake early warning system, the Internet of Things perception layer shoulders the important task of real-time collecting seismic wave data and completing preliminary signal processing and transmission. By deploying triaxial accelerometers, geomagnetic sensors, and temperature compensation units, it realizes the rapid perception of earthquake events, achieves multi-sensor collaborative acquisition, temperature compensation calibration, efficient signal conversion and transmission. With high-precision acquisition and low-latency transmission of seismic wave data, it lays a data foundation for the real-time performance and reliability of the earthquake early warning system, and also provides high-quality raw data for subsequent edge computing and cloud analysis. The Internet of Things perception layer is composed of low-power seismic monitoring sensors and data acquisition units. The low-power seismic monitoring sensors are deployed at building foundations, bridge nodes, and subway tunnels. The triaxial accelerometers are based on MEMS technology, which can capture ground three-dimensional vibration signals in real time and output digital signals. The geomagnetic sensors use the magnetoresistive effect to monitor the change in geomagnetic field intensity and propagation direction, and assist in identifying geomagnetic anomalies. The temperature compensation unit monitors the ambient temperature in real time through thermistors, corrects the sensor drift error according to the pre-stored calibration curve, and calibrates the interference of the ambient temperature on the sensor data to ensure the measurement accuracy. These components work together to detect ground vibration and geomagnetic field changes in real time, collect three-dimensional vibration signals and geomagnetic field data for identifying earthquake precursors. The analog-to-digital converter uses a successive approximation ADC, which can improve the sampling frequency, reduce the quantization error, and convert the analog signals output by the low-power seismic monitoring sensors into digital signals. Subsequently, the digital signals are transmitted to the edge computing layer through the Zigbee module using a low-power wireless protocol, and the Zigbee module supports self-organizing networks, which can expand the coverage range and achieve efficient data transmission.

[0021] The edge computing layer includes a noise filtering module, a waveform comparison module, and a local cache unit. The noise filtering module includes an input layer, a convolutional neural network, and an output layer. The waveform comparison module includes an amplitude comparison unit, a frequency comparison unit, and a period similarity unit. The local cache unit includes an FPGA chip and a pre-computation strategy library. As the local data processing core of the earthquake early warning system, the edge computing layer performs key tasks such as real-time noise filtering, waveform comparison, and strategy pre-computation at the edge node close to the data source. Its core objective is to reduce the cloud load and improve the system response speed through local processing, ensuring that the earthquake early warning can achieve second-level real-time performance. The edge computing layer receives the original seismic wave data transmitted from the Internet of Things perception layer, processes it through a series of operations, and then uploads the key data to the cloud platform to serve the subsequent global optimization analysis, realizing efficient, reliable, and low-latency earthquake early warning data processing, laying a solid foundation for cloud global optimization and emergency response. The noise filtering module uses a convolutional neural network model to remove noise from the original seismic waveform to improve the signal effectiveness. The input layer of this module receives the standardized waveform data. The convolutional neural network extracts the local features of the waveform through 3 convolutional kernels and uses the ReLU activation function to enhance the non-linear expression ability. The formula is: , is the original seismic wave waveform data transmitted from the Internet of Things perception layer, is the 3 convolutional kernels, is the rectified linear activation function, is the classifier, is the filtered effective seismic signal. The output layer uses the Softmax classifier to label the effective signal, reduce the mis-deletion rate, and output the filtered effective seismic signal for use by the waveform comparison module. The waveform comparison module confirms the earthquake event type and consistency through multi-dimensional comparison. Among them, the amplitude comparison unit calculates the amplitude difference between adjacent sensors to exclude local interference. The frequency comparison unit uses the fast Fourier transform to extract the main frequency. When the difference is less than 3Hz, it is determined as an earthquake event. The formula is: , is the frequency eigenvalue obtained after the FFT calculation of the seismic wave signal, is the original seismic wave time-domain data collected by the Internet of Things perception layer, is the number of data points per single sampling, determined by the sensor sampling frequency, and is the signal length for frequency-domain analysis, is the frequency index, corresponding to different frequency components, , is the imaginary unit, used for complex number operations to realize the conversion from time domain to frequency domain. The period similarity unit calculates the waveform cosine similarity through a sliding window to confirm the consistency of multi-sensor data. The formula is: , is the seismic waveform vector collected by adjacent sensors in the same sliding window, is the waveform vector at The amplitude / frequency value at each time point corresponds to the discrete sampling point of the seismic wave signal. is the data length of the sliding window, used for local waveform consistency analysis, It is the waveform similarity value, ranging from [-1,1]. The closer the absolute value is to 1, the more similar the waveform is. If all three comparisons are passed, it is determined to be a valid seismic wave and the early warning process is triggered. The local cache unit is responsible for storing the pre-calculated emergency strategy to ensure that the system can still respond quickly in the case of network disconnection. The FPGA chip relies on the hardware parallel computing capability to pre-store the elevator parking strategy and the seismic wave arrival time prediction model to optimize the local decision-making speed. The pre-calculated strategy library caches the emergency instructions in extreme scenarios. When the network is disconnected, the local strategy can be directly called to ensure that elevators, subways and other equipment complete the corresponding operations before the S wave arrives.

[0022] The cloud platform analysis module includes a machine learning model, a dynamic threshold adjustment unit and an encryption parameter generation unit. The machine learning model includes a support vector machine and a random forest model. The dynamic threshold adjustment unit includes a geological data interface, an amplitude threshold generator and a frequency threshold generator. The encryption parameter generation unit includes an SM3 module and an SM4 key generator. As the intelligent decision-making center of the earthquake early warning system, the cloud platform analysis module undertakes the important responsibilities of global data integration, dynamic threshold optimization and secure encryption. It uses a machine learning model to integrate and analyze local sensor data, earthquake monitoring center signals and geological database data, dynamically adjusts early warning parameters and generates encryption instructions to ensure that early warning information can be accurately and securely distributed to terminals. It receives pre-processed data uploaded by the edge computing layer, and simultaneously accesses the authoritative data of the earthquake monitoring center. After a series of processing, it pushes the global warning instructions to key facilities such as subways, elevators, and electricity, as well as public terminals through a multi-channel distribution module, realizing the integration of intelligent decision-making, dynamic optimization, and security encryption, providing core computing power and security for the earthquake early warning system. The machine learning model uses support vector machines and random forest models to classify earthquake types and predict earthquake intensity and impact range to improve the accuracy of early warning. The support vector machine constructs a classification hyperplane based on the radial basis kernel function, which can distinguish earthquake wave types with high accuracy. By learning and training historical earthquake data, it can identify abnormal waveforms. The formula is: ,in, is the seismic wave eigenvector, are seismic wave samples to be classified, each of which contains multiple multidimensional features. is the feature vector of the sample, is the amplitude, is the main frequency, is the waveform period, is the kernel function parameter that controls the "tightness" of the feature space mapping. It is determined by training with historical earthquake data and is the square of the Euclidean distance between two sample feature vectors, used to measure feature differences. The random forest model integrates 100 decision trees and judges the earthquake intensity and propagation range by voting. It incorporates multi-source data, including sensor frequency, waveform period, fault distribution, soil stiffness coefficient, regional geological classification, epicenter location, magnitude, focal depth, historical earthquake waveforms, earthquake influence range, and a real-time adjusted frequency threshold. The formula is: , is the input sample, including sensor data and geological parameters. is the th decision tree's prediction result for the sample . is the predefined earthquake intensity level or influence range category. is the indicator function. is the earthquake intensity / influence range finally predicted by the random forest. The dynamic threshold adjustment unit dynamically optimizes the warning threshold according to geological conditions to reduce the misjudgment rate. The geological data interface accesses the GIS geographic information system to obtain fault distribution and soil stiffness coefficient data, and maps these geological data to the warning area after parsing. The amplitude threshold generator sets a dynamic amplitude range according to different geological conditions, and the frequency threshold generator filters out non-seismic frequency band interference for different geological conditions to make the warning threshold adapt to different geological conditions. The encryption parameter generation unit is committed to ensuring the security of data transmission, preventing information from being tampered with and stolen. The SM3 module compresses the data using the hash algorithm to generate a 256-bit digest to ensure data integrity and can adapt to high-concurrency scenarios. The SM4 key generator dynamically generates a 256-bit symmetric encryption key, and combines with the blockchain smart contract deployed based on the Ethereum private chain to synchronize the key to the edge node in the light node mode to form a blockchain network, greatly improving the key synchronization success rate. In the case of network disconnection, the preset key library is enabled to store the keys in the recent 24 hours to ensure the continuity of encryption.

[0023] The multi-channel distribution module includes a communication unit and an offline device wake-up unit. The communication unit includes a Beidou transceiver, a 5G interface, and a digital broadcast transmitter. The offline device wake-up unit includes a Beidou instruction encoder and rural emergency broadcast terminals. As the information transmission center of the earthquake early warning system, the multi-channel distribution module ensures that early warning information can reach different scenarios quickly and reliably by building redundant communication paths. It reasonably allocates communication resources according to the priority of the recipients and has the ability to wake up emergency broadcast terminals in areas without network through Beidou satellite instructions. After receiving the early warning instructions sent by the cloud platform, this module distributes them to different terminals according to the priority, realizing high-reliability, full-scenario, and low-latency transmission of early warning information, and building a communication guarantee for earthquake emergency response. The communication unit transmits early warning information through multiple channels, aiming to achieve full coverage of urban and remote areas. The Beidou transceiver sends short messages containing encrypted instructions based on the RDSS protocol to cover areas without network and effectively prevent information from being tampered with. The 5G interface adopts the NSA / SA dual-mode networking mode, allocating exclusive bandwidth for key facilities with the first-level priority such as subways and hospitals, which can not only quickly push early warning instructions but also support real-time video linkage. The digital broadcast transmitter sends evacuation guidelines through FM radio, and can synchronously push voice broadcasts and text information to transmit early warning content to areas with the second-level priority such as schools and shopping malls. The offline device wake-up unit is mainly responsible for waking up rural emergency broadcast terminals without network connection. After encoding the wake-up instructions into BCD format by the Beidou instruction encoder, they are sent to rural emergency broadcast terminals through Beidou satellites. These terminals support the offline working mode and are equipped with built-in lithium batteries, widely covering rural administrative villages and natural villages. Once receiving the Beidou satellite instructions, the terminals will automatically broadcast the pre-stored voice warnings to ensure that the rural areas, that is, the areas with the third-level priority, can also receive early warning information in a timely manner. When the main network fails, the multi-channel distribution module can automatically switch to the backup channel and trigger the offline device wake-up function to ensure the stability of early warning information transmission.

[0024] The dynamic fault tolerance module includes a network status monitoring unit, a redundancy check unit, and a local emergency instruction library. The network status monitoring unit includes a packet loss rate detector and a latency detector. The redundancy check unit includes a CRC32 checker and a data retransmission controller. The local emergency instruction library includes an elevator docking plan library and a subway braking instruction library. As the core for ensuring the reliability of the earthquake early warning system, the dynamic fault tolerance module maintains the continuity of system functions in case of network anomalies or hardware failures. It ensures the accurate distribution and execution of early warning signals under extreme conditions by monitoring the network status in real time, using redundancy checks to ensure data integrity, and invoking locally stored instructions, achieving a three-in-one fault tolerance ability of network self-healing, data reliability, and offline response, providing all-weather and full-scenario stability guarantee for the earthquake early warning system. The network status monitoring unit uses distributed probe technology to collect network quality indicators in real time and continuously monitor the network communication status to determine whether to trigger the fault tolerance mechanism. This unit uses the packet loss rate detector to count the proportion of lost data packets and issues an alarm once the trigger threshold is reached to identify network congestion or link interruption. At the same time, the latency detector measures the round-trip time of data packets to determine the network response ability. When a high-latency scenario is detected, it timely triggers the communication channel switch. By continuously counting the packet loss situation during data transmission and combining with a dynamic threshold model to judge the network health, measuring the round-trip time of data transmission to identify network congestion or interruption, and timely triggering the switch between the main network and the backup communication channel according to the detection results. The redundancy check unit adopts a hierarchical check mechanism to ensure the integrity and reliability of data transmission. The CRC32 checker uses cyclic redundancy check codes to verify the received data, filtering out transmitted errors or damaged data to ensure the integrity of data packets. Once the check fails, the data retransmission controller automatically retransmits the data packet until it is successfully received or the maximum retransmission times are reached, improving the delivery reliability of critical instructions, excluding errors during transmission, and ensuring that the instruction transmission is not tampered with or lost. The local emergency instruction library stores pre-computed emergency instructions to ensure that the system can still execute critical operations when the network is completely interrupted. The elevator docking plan library stores the corresponding docking logic and priorities for different floor heights and can be directly invoked during network disconnection to drive the elevator to dock safely. The subway braking instruction library caches the train braking parameters and triggers the emergency braking of the hydraulic system during network interruption to prevent the train from derailing. Through the version synchronization mechanism, the local emergency instruction library pre-stores the cloud optimization strategy locally to ensure the continuity of emergency operations in the offline scenario.

[0025] The emergency linkage module includes an elevator intelligent docking unit, a subway braking unit, and a power cut-off unit. The elevator intelligent docking unit includes a laser rangefinder and an elevator controller. The subway braking unit includes a braking signal generator and a hydraulic braking actuator. The power cut-off unit includes an intelligent circuit breaker and a power status feedback module. As the terminal execution center of the earthquake early warning system, the emergency linkage module shoulders the important task of accurately converting early warning instructions into equipment control actions. By linking key facilities such as elevators, subways, and power, it aims to minimize the risk of earthquake disasters. It receives early warning instructions issued by the multi-channel distribution module, assigns execution priorities according to preset rules, sends control signals to terminal devices via a standardized protocol, and monitors the execution status in real time. Once the device response times out or the execution fails, it automatically triggers an alternative plan. With standardized protocol integration and intelligent priority scheduling, it ensures that the earthquake early warning forms a seamless closed loop from "information perception" to "terminal execution", significantly enhancing the seismic resilience of the urban lifeline system. The elevator intelligent docking unit is committed to ensuring that the elevator safely docks to the designated floor before the arrival of seismic waves to prevent passengers from being trapped. It monitors the elevator position and operating status in real time, generates a docking strategy based on the predicted arrival time of seismic waves, and then drives the elevator control system to execute the operation. The laser rangefinder uses time-of-flight technology to accurately measure the distance between the elevator car and the floor to determine whether the elevator is between floors. If it is between floors, it triggers the emergency docking logic. After receiving the docking instruction, if the elevator is between floors, the elevator controller will control the motor to drive the car to move smoothly and quickly to the nearest safe floor, then lock the access control and cut off the operating power to avoid the risk of secondary shaking;If the elevator has stopped, directly lock the elevator doors and cut off the operating power supply. At the same time, feedback the stopping result to the cloud platform to update the floor evacuation map. The subway braking unit is responsible for triggering the emergency braking of the subway train to prevent derailment or collision accidents. This unit predicts the arrival time of the seismic wave based on the propagation speed of the seismic wave and the current position of the train, and then calculates the braking distance and pressure parameters to generate a hierarchical braking instruction, and hierarchically executes deceleration or emergency braking operations to drive the hydraulic system to apply braking force. At the same time, synchronously monitor the temperature and pressure status of the brake disc and link the platform public address system to release evacuation information. The braking signal generator generates a PWM signal to accurately control the starting time and strength of the braking system, avoiding passenger injuries or equipment damage caused by sudden braking. The hydraulic braking actuator applies braking force through the hydraulic cylinder, and monitors the temperature and pressure status of the brake disc in real time to ensure the smooth and effective braking process and prevent the occurrence of brake failure. The main function of the power cut-off unit is to cut off the power supply in high-risk areas to prevent electrical fires or equipment short circuits. Based on the GIS map, it can locate high-risk facilities near the epicenter, generate a list of power cut-off instructions, and remotely control the circuit breaker to trip, giving priority to cutting off non-essential loads while retaining the emergency power supply, and real-time feedback the power grid break point information to the command center to provide support for the rapid restoration of power supply after the disaster. After receiving the tripping instruction, the intelligent circuit breaker triggers the power cut-off operation by the electromagnetic release. It supports remote reset and status feedback functions, can accurately cut off the target circuit, and retain the emergency power supply line. The power status feedback module monitors the grid voltage and current parameters through sensors and encrypts and transmits these data back to the command center to provide real-time data basis for the restoration of power supply after the disaster.;

[0026] The self-check and fault recovery module includes a health monitoring unit, a fault switching and recovery unit, and a log and feedback unit. As the health management center of the earthquake early warning system, the self-check and fault recovery module achieves real-time perception, automatic switching, and recovery of faults by periodically detecting the operating status of the system's hardware and software, ensuring the continuous and reliable operation of the system under abnormal conditions such as equipment failures and network interruptions. Through periodic self-checks, intelligent switching, and closed-loop feedback, it builds a full-link resilience ability from "fault perception" to "rapid recovery" for the earthquake early warning system, significantly enhancing the system's robustness in complex environments. The health monitoring unit detects the operating status of system components in real time through heartbeat signals and data consistency verification, identifies potential faults, determines whether sensors are malfunctioning, simultaneously detects battery power and temperature drift, issues warnings for low battery or environmental interference, counts network packet loss rates and delays, identifies communication link congestion or interruptions, monitors CPU and memory usage, and prevents data processing delays caused by hardware overload, comprehensively ensuring the stable operating status monitoring of the system's core components. The fault switching and recovery unit automatically switches to backup resources to restore system functions when a fault occurs. When a sensor fails, it switches to a backup sensor in the same area to maintain the continuity of data collection. When the main communication module fails, it enables the Beidou satellite or local broadcast link to ensure uninterrupted communication. When a process crash is detected, it automatically restarts the service and loads the most recent cached status. When a memory leak occurs, it releases resources and triggers an alarm to notify the operations and maintenance personnel, quickly responding to various faults through these measures and restoring the normal operation of the system. The log and feedback unit is responsible for recording fault information and feeding it back to the cloud to assist in system optimization. It encrypts and stores data such as fault time, type, and recovery measures, facilitating root cause analysis afterwards. Based on historical fault frequencies and recovery efficiencies, it dynamically evaluates the health levels of devices and links, and combines with the machine learning model of the cloud platform to push maintenance suggestions, providing strong data support and decision-making basis for the continuous optimization of the system.

[0027] As the core of information display and user access in the earthquake early warning system, the user interface undertakes the important task of intuitively and efficiently transmitting early warning information, evacuation guidance, and system status to end users. At the same time, it supports user feedback and interactive operations. By receiving real-time early warning data sent from the cloud platform, it generates visual early warning information in combination with the GIS map, and dynamically updates evacuation routes, safe areas, and equipment status. It adapts to different user scenarios through various methods such as voice announcements, vibration reminders, and graphic and text push, collects feedback from users on confirming receipt of instructions and transmits it back to the cloud platform for optimizing emergency strategies. With the design of the trinity of dynamic visualization, multi-channel alerts, and intelligent feedback, it constructs a complete link from early warning perception to user actions, significantly improving the public's emergency response ability and system transparency. It integrates GIS geographical information and real-time earthquake data to generate a visual early warning map, intuitively showing the earthquake impact range, evacuation routes, and the status of key facilities, capable of rendering the epicenter location, intensity distribution, and safe areas, dynamically marking evacuation routes, shelters, and risk areas, and supporting various interactive operations. It can be adapted to the display of large-screen command centers and mobile terminals, facilitating users to obtain accurate information from different devices. By means of multiple channels, it ensures that users receive early warning information in a timely manner. In public places, it automatically switches the multi-language evacuation guidance broadcast by the public address system according to the regional population distribution, personal devices receive voice prompts, and volume adaptive adjustment is supported. Mobile terminals prompt emergency alerts through the vibration mode, and the vibration intensity is dynamically adjusted according to the earthquake intensity. At the same time, it pushes graphic and text cards containing key information, and the cards support one-key navigation to the nearest shelter, comprehensively ensuring that users can effectively receive early warnings in different scenarios. It collects user behavior data to optimize the system interaction experience. Users can report their status through the interface buttons, and the data is encrypted and transmitted back to the command center. Voice command input is supported, and emergency requests are generated through NLP parsing. Users can also customize their alert preferences, and the configured information can be synchronized in the cloud. In addition, an accessibility mode is provided to adapt to the needs of elderly and visually impaired users, fully considering the usage experience of different user groups. It displays the system operation status and equipment health for decision-making reference by operation and maintenance personnel. It integrates an equipment status dashboard, uses red, yellow, and green colors to identify the health levels, shows the execution progress of emergency operations, and supports manual intervention, facilitating operation and maintenance personnel to promptly understand the system situation and take corresponding measures.

[0028] After the edge computing layer receives the original seismic wave data transmitted by the Internet of Things perception layer, it first preprocesses the data through a noise filtering module, uses a convolutional neural network model to identify and eliminate environmental noise, and retains the effective seismic signals. Subsequently, the waveform comparison module performs multi-dimensional analysis on the filtered data, including amplitude difference, frequency difference, and waveform period similarity comparison, to confirm the consistency of the seismic wave type and the event. After local processing, the edge computing layer uploads the verification results to the cloud platform analysis module to provide data support for global early warning decision-making. After receiving the comparison results uploaded by the edge computing layer, the cloud platform analysis module fuses multi-source data through a machine learning model, dynamically adjusts the early warning threshold, and generates global early warning instructions. After being encrypted by the encryption parameter generation unit, the instructions are sent to the multi-channel distribution module. According to the preset priority strategy, the multi-channel distribution module pushes the early warning instructions to different terminals: the first-priority instructions are transmitted in real time through the 5G private network; the second-priority instructions are released through digital broadcasting; the third-priority instructions reach offline devices through Beidou satellites to ensure full-scenario coverage. The dynamic fault tolerance module monitors the network communication status in real time. When it detects that the main network packet loss rate or delay exceeds the limit, it automatically switches to Beidou satellites or the local emergency instruction library. The redundancy check unit performs integrity verification on the transmitted data. If the verification fails, it triggers an automatic retransmission mechanism. When the network is completely interrupted, the local emergency instruction library directly calls the pre-stored strategy to drive the emergency linkage module to execute operations. After receiving the instructions, the emergency linkage module triggers the intelligent docking of elevators, hierarchical braking of subways, and power cut-off in high-risk areas, and encrypts and feeds back the execution status to the cloud platform to form a closed-loop disaster tolerance system.

[0029] The system operation process is as follows: S1. Data acquisition and preliminary processing S1.1 Sensor network startup: Low-power seismic monitoring sensors deployed on building foundations, bridges, and subway tunnels collect ground vibration, geomagnetic field changes, and environmental temperature data in real time; The sensor eliminates environmental interference through the temperature compensation unit to ensure data accuracy; S1.2 Signal conversion and transmission: The data acquisition unit converts the analog signal output by the sensor into a digital signal through an analog-to-digital converter, and the digital signal is transmitted to the edge computing layer through the Zigbee module using a low-power wireless protocol, supporting self-organizing network to expand the coverage range; S2. Edge computing layer real-time processing S2.1 Noise filtering: The noise filtering module of the edge computing layer uses a convolutional neural network to eliminate noises such as traffic vibration and mechanical interference, and retains the effective seismic signals; S2.2 Waveform comparison and confirmation: The waveform comparison module performs three comparisons: Amplitude comparison: Calculate the amplitude difference between adjacent sensors to exclude local interference; Frequency comparison: Extract the main frequency through the fast Fourier transform to determine whether it is an earthquake event; Period similarity comparison: Calculate the waveform cosine similarity to confirm the consistency of multi-sensor data; If all three comparisons pass, it is determined as an effective seismic wave; S2.3 Local policy pre-storage: The local cache unit stores the pre-computed emergency policies through the FPGA chip and directly calls them when the network is disconnected; S2.4 Data upload: The processed key data is uploaded to the cloud platform analysis module; S3. Global optimization and instruction generation on the cloud platform S3.1 Multi-source data fusion: The machine learning model fuses sensor data, geological database, and historical earthquake data to classify earthquake types and predict the intensity and impact range; S3.2 Dynamic threshold adjustment: Dynamically optimize the amplitude and frequency warning thresholds according to geological conditions to reduce the false alarm rate; S3.3 Encryption and instruction generation: The encryption parameter generation unit encrypts data using the national cryptography SM3 / SM4 algorithm and synchronizes the encryption keys to the edge nodes in combination with blockchain technology; Generate global warning instructions; S4. Multi-channel redundant distribution S4.1 Priority distribution strategy: The communication unit pushes instructions according to priorities: First-level priority: Real-time transmission through the 5G private network; Second-level priority: Release evacuation guidelines through digital broadcasting; Third-level priority: Send encrypted short messages through Beidou satellites to wake up the emergency broadcast terminal; S4.2 Wake-up of offline devices: The Beidou instruction encoder sends specific encoded instructions to wake up the rural emergency broadcast terminal and broadcast the pre-stored voice warning; S5. Dynamic fault tolerance and emergency execution S5.1 Network anomaly handling: The dynamic fault tolerance module monitors the network packet loss rate and latency in real time: When the main network is abnormal, automatically switch to the Beidou satellite or local broadcast link; The redundancy check unit ensures the integrity of the instructions through CRC32 check and automatic retransmission; S5.2 Emergency response in case of network disconnection: The local emergency instruction library calls the pre-stored strategy to directly trigger the operations of elevators, subways, and power equipment; S5.3 Terminal device linkage: The emergency linkage module executes operations after receiving instructions: Intelligent elevator docking: The laser rangefinder locates the elevator position and docks it to a safe floor; Subway hierarchical braking: Generate hierarchical braking signals to complete emergency braking; Power cut-off: Cut off the power supply in high-risk areas and retain the emergency power supply for hospitals and subways; S6. User Interaction and Feedback Closed-loop S6.1 Multi-modal Warning Reach: The user interaction interface publishes warning information to the public through voice announcements, vibration reminders, and graphic and text push, and supports one-key navigation to shelters; S6.2 Visual Map Display: Dynamically render the epicenter location, evacuation routes, and safe areas, and adapt to large-screen command centers and mobile terminals; S6.3 User Feedback Collection: Users report their status through interface buttons or voice commands, and the data is encrypted and transmitted back to the cloud platform to optimize the strategy; S7. System Health Management and Maintenance S7.1 Periodic Self-check: The self-check and fault recovery module detects the status of sensors, networks, and hardware, and warns of low battery or overload; S7.2 Automatic Fault Switching: Switch to a standby node when the sensor fails, and automatically restart the service when the process crashes; S7.3 Logging and Optimization: Encrypt and store fault logs, and push maintenance suggestions in combination with machine learning models.

[0030] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0031] The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. Internet of Things multi-channel earthquake early warning emergency linkage system, characterized in that: It includes an Internet of Things perception layer, an edge computing layer, a cloud platform analysis module, a multi-channel distribution module, a dynamic fault-tolerant module, an emergency linkage module, a self-check and fault recovery module and a user interaction interface. The Internet of Things perception layer is responsible for the collection and preliminary processing of seismic wave data. The edge computing layer performs noise filtering, waveform comparison and strategy pre-calculation locally to reduce cloud load. The cloud platform analysis module achieves global optimization through machine learning and dynamic threshold adjustment, and generates encryption parameters. The multi-channel distribution module ensures that the warning information reaches different scenarios through multi-path communication. The dynamic fault-tolerant module monitors the network status and performs redundancy verification to ensure the reliability of the system under abnormal conditions. The emergency linkage module controls elevators, subways and power equipment according to the warning instructions to achieve rapid and safe response. The self-check and fault recovery module includes a health monitoring unit, a fault switching and recovery unit and a log and feedback unit. The self-check and fault recovery module automatically detects the health status of the sensor every 24 hours and switches to the backup sensor when a fault occurs. The user interaction interface provides a visual warning map and emergency guidance, and supports voice broadcast, vibration reminder and multi-language switching functions.

2. The Internet of Things multi-channel earthquake early warning emergency linkage system as claimed in claim 1, characterized in that: The Internet of Things perception layer includes a low-power earthquake monitoring sensor and a data acquisition unit. The low-power earthquake monitoring sensor includes a three-axis accelerometer, a geomagnetic sensor and a temperature compensation unit. The data acquisition unit includes a Zigbee module and an analog-to-digital converter.

3. The Internet of Things multi-channel earthquake early warning emergency linkage system as claimed in claim 1, characterized in that: The edge computing layer includes a noise filtering module, a waveform comparison module and a local cache unit. The noise filtering module includes an input layer, a convolutional neural network and an output layer. The waveform comparison module includes an amplitude comparison unit, a frequency comparison unit and a period similarity unit. The local cache unit includes an FPGA chip and a pre-calculation strategy library.

4. The Internet of Things multi-channel earthquake early warning emergency linkage system as claimed in claim 1, characterized in that: The cloud platform analysis module includes a machine learning model, a dynamic threshold adjustment unit and an encryption parameter generation unit. The machine learning model includes a support vector machine and a random forest model. The dynamic threshold adjustment unit includes a geological data interface, an amplitude threshold generator and a frequency threshold generator. The encryption parameter generation unit includes an SM3 module and an SM4 key generator.

5. The Internet of Things multi-channel earthquake early warning emergency linkage system as claimed in claim 4, characterized in that: The multi-channel distribution module includes a communication unit and an offline device wake-up unit. The communication unit includes a Beidou transceiver, a 5G interface and a digital broadcast transmitter. The offline device wake-up unit includes a Beidou command encoder and a rural emergency broadcast terminal.

6. The Internet of Things multi-channel earthquake early warning emergency linkage system as claimed in claim 1, characterized in that: The dynamic fault-tolerant module includes a network status monitoring unit, a redundancy check unit and a local emergency instruction library. The network status monitoring unit includes a packet loss rate detector and a delay detector. The redundancy check unit includes a CRC32 checker and a data retransmission controller. The local emergency instruction library includes an elevator stop plan library and a subway braking instruction library.

7. The Internet of Things multi-channel earthquake early warning emergency linkage system as claimed in claim 1, characterized in that: The emergency linkage module includes an elevator intelligent parking unit, a subway braking unit and a power cut-off unit. The elevator intelligent parking unit includes a laser rangefinder and an elevator controller. The subway braking unit includes a braking signal generator and a hydraulic brake actuator. The power cut-off unit includes an intelligent circuit breaker and a power status feedback module.

8. The Internet of Things multi-channel earthquake early warning emergency linkage system as claimed in claim 1, characterized in that: The system operation process is as follows: S1. Data collection and preliminary processing S1.1 Sensor network startup: Low-power earthquake monitoring sensors deployed on building foundations, bridges, and subway tunnels collect ground vibration, geomagnetic field changes, and ambient temperature data in real time; The sensor eliminates environmental interference through the temperature compensation unit to ensure data accuracy; S1.2 Signal conversion and transmission: The data acquisition unit converts the analog signal output by the sensor into a digital signal through an analog-to-digital converter. The digital signal is transmitted to the edge computing layer through the Zigbee module using a low-power wireless protocol, supporting self-organizing networks to expand coverage. S2. Real-time processing at the edge computing layer S2.1 Noise filtering: The noise filtering module of the edge computing layer uses a convolutional neural network to remove noise from traffic vibration and mechanical interference, retaining valid seismic signals; S2.2 Waveform comparison and confirmation: The waveform comparison module performs three comparisons: Amplitude comparison: calculate the amplitude difference between adjacent sensors to eliminate local interference; Frequency comparison: extract the main frequency through fast Fourier transform to determine whether it is an earthquake event; Cycle similarity comparison: calculate the waveform cosine similarity and confirm the consistency of multi-sensor data; If all three comparisons are passed, it is determined to be a valid seismic wave; S2.3 Local strategy pre-storage: The local cache unit stores the pre-calculated emergency strategy through the FPGA chip and directly calls it when the network is disconnected; S2.4 Data upload: The processed key data is uploaded to the cloud platform analysis module; S3. Cloud platform global optimization and instruction generation S3.1 Multi-source data fusion: The machine learning model integrates sensor data, geological databases, and historical earthquake data to classify earthquake types and predict intensity and impact range; S3.2 Dynamic threshold adjustment: Dynamically optimize amplitude and frequency warning thresholds according to geological conditions to reduce misjudgment rate; S3.3 Encryption and instruction generation: The encryption parameter generation unit uses the national encryption SM3 / SM4 algorithm to encrypt data and synchronizes the key to the edge node in combination with blockchain technology; Generate global warning instructions; S4. Multi-channel redundant distribution S4.1 Priority distribution strategy: The communication unit pushes instructions according to priority: Level 1 priority: real-time transmission through 5G private network; Priority Level 2: Evacuation instructions issued via digital broadcasting; Level 3 priority: Send encrypted short messages via Beidou satellites to wake up emergency broadcast terminals; S4.2 Offline device wake-up: Beidou command encoder sends coded commands to wake up the rural emergency broadcast terminal and broadcast pre-stored voice warnings; S5. Dynamic fault tolerance and emergency execution S5.1 Network exception handling: Dynamic fault tolerance module monitors network packet loss rate and delay in real time: When the main network is abnormal, it automatically switches to the Beidou satellite or local broadcast link; The redundancy check unit ensures command integrity through CRC32 check and automatic retransmission; S5.2 Network disconnection emergency response: The local emergency command library calls pre-stored strategies to directly trigger elevator, subway, and power equipment operations; S5.3 Terminal equipment linkage: The emergency linkage module performs operations after receiving the command: Intelligent elevator parking: The laser rangefinder locates the elevator and stops it at a safe floor; Subway graded braking: Generate graded braking signals to complete emergency braking; Power cut-off: Cut off power supply to high-risk areas, and retain emergency power supply to hospitals and subways; S6. User interaction and feedback loop S6.1 Multi-modal warning access: The user interface releases warning information to the public through voice broadcast, vibration reminder, and text and image push, and supports one-click navigation to the shelter; S6.2 Visual map display: Dynamic rendering of epicenter location, evacuation routes and safe areas, adapted to large-screen command centers and mobile terminals; S6.3 User feedback collection: Users report status through interface buttons or voice commands, and the data is encrypted and sent back to the cloud platform for optimization strategy; S7. System health management and maintenance S7.1 Periodic self-check: The self-check and fault recovery module detects the status of sensors, networks, and hardware, and warns of low battery or overload; S7.2 Automatic failover: switch to the backup node when the sensor fails, and automatically restart the service when the process crashes; S7.3 Logging and Optimization: Encrypted storage of fault logs, and push maintenance recommendations based on machine learning models.

Images