An automatic monitoring system and method for civil air defense door

Abstract

The application discloses a kind of civil air defence door automatic monitoring system and method, belong to civil air defence engineering technical field.The system includes state perception module, edge calculation module, state evaluation module, early warning decision module and execution feedback module: state perception module gathers multiple state data;Edge calculation module generates standardization feature vector sequence after receiving data;State evaluation module is based on structure health degree, opening and closing function integrity, environmental safety degree three sub-models weighted fusion, calculates civil air defence door comprehensive state index and generates state evaluation report;Early warning decision module determines early warning level according to multilevel threshold rule and outputs corresponding control instruction;Execution feedback module drives door body execution mechanism to execute action, and execution result is fed back to state perception module, forms complete monitoring closed loop.The application can realize the multidimensional real-time monitoring and intelligent closed-loop control of civil air defence door state, effectively improve the progressive deterioration early warning accuracy and emergency response reliability of sudden scene.

Description

Technical Field

[0001] This invention relates to the field of civil defense engineering technology, specifically to an automatic monitoring system and method for civil defense doors. Background Technology

[0002] Civil defense doors are key protective equipment in civil air defense projects, widely used in entrances and exits of underground civil defense projects, subway partition doors, and protective doorways for important infrastructure. Their core function is to effectively block threats such as blast waves, toxic gases, and flames during wartime or sudden disasters, ensuring the safety of personnel and equipment. With the rapid development of sensor technology, IoT technology, and edge computing technology, the status monitoring of civil defense doors has gradually evolved from traditional manual periodic inspections to automation and intelligence. Existing technologies include various sensor-based civil defense door status monitoring solutions, such as using attitude sensors to monitor the door's hanging angle, strain sensors to monitor the tension of the tie rods, and temperature, humidity, and gas sensors to monitor surrounding environmental parameters. Some solutions also integrate video surveillance and image recognition technology to achieve remote detection of surface deformation and corrosion on civil defense doors.

[0003] However, existing technical solutions have the following objective technical defects: First, existing monitoring systems mostly use single-type sensors or simple sensor combinations, lacking the ability to synchronously perceive multi-dimensional stress on the door structure, opening and closing function status, and environmental parameters, making it difficult to comprehensively reflect the overall operating status of the air-raid shelter door; Second, the data processing links of existing solutions mostly rely on remote cloud servers, resulting in long data transmission links and high latency. In underground enclosed environments, when network signals are unstable, data loss or processing delays are likely to occur, failing to meet the timeliness requirements of real-time monitoring of the air-raid shelter door status; Third, existing assessment methods mostly use fixed threshold judgments or simple rule judgments, without establishing a comprehensive status assessment model based on multi-source data fusion, resulting in insufficient early warning capabilities for progressive door degradation and multi-factor coupled failures; Fourth, the monitoring and execution control links of existing systems are independent of each other, lacking a closed-loop feedback mechanism between status monitoring results and door execution mechanisms, failing to automatically trigger protective actions when abnormal states are detected, thus reducing the system's emergency response efficiency. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an automatic monitoring system and method for air defense doors, which can synchronously perceive the structure, opening and closing function and surrounding environment of air defense doors in multiple dimensions, ensure the real-time performance and data integrity of status assessment in underground weak network environment, provide quantitative early warning of progressive deterioration of door body, and automatically complete the execution and verification of protective actions in abnormal state.

[0005] To address the aforementioned technical problems, this invention provides an automatic monitoring system for air-raid shelter doors, comprising:

[0006] The state perception module is used to collect multi-dimensional state data of the air defense door body and the surrounding environment. The multi-dimensional state data includes door structure deformation data, door opening and closing position data, ambient temperature and humidity data, ambient gas concentration data, and door vibration spectrum data.

[0007] The edge computing module is communicatively connected to the state awareness module and is used to receive the multi-dimensional state data, perform local data preprocessing, multi-source data spatiotemporal alignment and feature extraction, and generate a standardized feature vector sequence.

[0008] The status assessment module is communicatively connected to the edge computing module. It is used to receive the standardized feature vector sequence, calculate the comprehensive status index of the air defense door based on the preset multidimensional status assessment model, and generate a status assessment report.

[0009] The early warning decision module is communicatively connected to the status assessment module. It is used to receive the status assessment report, generate an early warning level determination result according to the preset multi-level threshold rules, and output the corresponding early warning control command.

[0010] The execution feedback module is communicatively connected to the early warning decision module. It is used to receive the early warning control command, drive the actuator of the air defense door to perform corresponding actions, and feed back the execution result to the status perception module to form a monitoring closed loop.

[0011] Preferably, the state perception module includes: a structural deformation perception unit, configured to deploy a strain sensor array at key load-bearing parts of the door body to collect stress distribution data and deformation displacement data of the door body structure;

[0012] The opening and closing state sensing unit is configured to install angle encoders and displacement sensors at the door hinge shaft and locking mechanism to collect door opening angle data and locking engagement depth data.

[0013] The environmental parameter sensing unit is configured to deploy temperature and humidity sensors, gas concentration sensors and vibration acceleration sensors around the door frame to collect ambient temperature data, ambient humidity data, harmful gas concentration data and door vibration spectrum data.

[0014] The data aggregation unit is communicatively connected to the structural deformation sensing unit, the opening / closing state sensing unit, and the environmental parameter sensing unit, respectively. It is used to aggregate and package the raw data collected by each sensing unit according to a unified timestamp and transmit it to the edge computing module.

[0015] Preferably, the edge computing module includes:

[0016] The data preprocessing unit is used to perform outlier removal, missing value imputation, and dimension normalization on the received raw data.

[0017] The spatiotemporal alignment unit is used to align timestamps of multi-source heterogeneous data based on a unified time reference, and to align the spatial positions of data from different measurement points of the door based on spatial coordinate mapping.

[0018] The feature extraction unit is used to extract time-domain statistical features, frequency-domain energy features, and time-frequency joint features from the aligned data to generate the standardized feature vector sequence.

[0019] The data preprocessing unit, the spatiotemporal alignment unit, and the feature extraction unit are connected in series, and the output of the feature extraction unit is communicatively connected to the input of the state evaluation module.

[0020] Preferably, the state assessment module includes:

[0021] An evaluation model storage unit is used to store the multi-dimensional state evaluation model, which includes a door structure health sub-model, an opening and closing function integrity sub-model, and an environmental safety sub-model trained based on historical monitoring data.

[0022] The comprehensive index calculation unit is used to input the standardized feature vector sequence into each sub-model to obtain the status score of each dimension, and perform weighted fusion based on preset weight coefficients to calculate the comprehensive status index of the civil defense door.

[0023] The report generation unit is used to generate a status assessment report containing status level identifiers, a list of abnormal parameters, and trend prediction information based on the comprehensive status index of the air defense door and the status scores of each dimension.

[0024] The evaluation model storage unit, the comprehensive index calculation unit, and the report generation unit are connected in sequence, and the output end of the report generation unit is communicatively connected to the input end of the early warning decision module.

[0025] On the other hand, the present invention provides an automatic monitoring method for air defense doors, comprising the following steps:

[0026] S1: State perception step, in which the state perception module collects multi-dimensional state data of the air defense door body and the surrounding environment. The multi-dimensional state data includes door structure deformation data, door opening and closing position data, ambient temperature and humidity data, ambient gas concentration data and door vibration spectrum data.

[0027] S2: Edge computing step, where the edge computing module receives the multi-dimensional state data, performs local data preprocessing, multi-source data spatiotemporal alignment and feature extraction, and generates a standardized feature vector sequence;

[0028] S3: Status assessment step, in which the status assessment module receives the standardized feature vector sequence, calculates the comprehensive status index of the air defense door based on the preset multi-dimensional status assessment model, and generates a status assessment report;

[0029] S4: Early warning decision-making step, in which the early warning decision-making module receives the status assessment report, generates an early warning level determination result according to the preset multi-level threshold rules, and outputs the corresponding early warning control command;

[0030] S5: Execution feedback step, whereby the execution feedback module receives the early warning control command, drives the personnel defense door actuator to perform the corresponding action, and feeds back the execution result to the status perception module to form a monitoring closed loop.

[0031] Preferably, the S1 state sensing step includes: the structural deformation sensing unit collecting stress distribution data and deformation displacement data of the door structure at key load-bearing parts of the door body;

[0032] The opening and closing status sensing unit collects door opening angle data and locking engagement depth data at the door hinge axis and locking mechanism.

[0033] The environmental parameter sensing unit collects ambient temperature data, ambient humidity data, harmful gas concentration data, and door vibration spectrum data around the door frame;

[0034] The data aggregation unit aggregates and packages the raw data collected by each sensing unit according to a unified timestamp, and then transmits it to the edge computing module.

[0035] Preferably, the S2 edge computing step includes: the data preprocessing unit performs outlier removal, missing value imputation, and dimension normalization on the received raw data;

[0036] The spatiotemporal alignment unit performs timestamp alignment on multi-source heterogeneous data based on a unified time reference, and performs spatial position alignment on data from different measuring points of the gate based on spatial coordinate mapping;

[0037] The feature extraction unit extracts time-domain statistical features, frequency-domain energy features, and time-frequency joint features from the aligned data to generate the standardized feature vector sequence.

[0038] Preferably, the S3 state assessment step includes: the assessment model storage unit provides the multi-dimensional state assessment model, including a door structure health sub-model, an opening and closing function integrity sub-model, and an environmental safety sub-model trained based on historical monitoring data;

[0039] The standardized feature vector sequence is input into each sub-model by the comprehensive index calculation unit to obtain the status score of each dimension, and is weighted and fused based on the preset weight coefficient to calculate the comprehensive status index of the civil defense door.

[0040] The report generation unit generates a status assessment report containing status level identifiers, a list of abnormal parameters, and trend prediction information based on the comprehensive status index of the air defense door and the status scores of each dimension.

[0041] The system described in this invention can be applied to real-time status monitoring scenarios of entrances and exits of underground civil defense projects, civil defense partition doors in subway sections, or protective doors of important infrastructure. The deployment environment is an underground enclosed space or a semi-enclosed passage, and the sensor placement positions of the status perception module are configured differently according to the mechanical characteristics of the door structure and the characteristics of the environmental airflow distribution.

[0042] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the above-described automatic monitoring method for air-raid shelter doors are implemented.

[0043] Compared with existing technologies, the beneficial effects of this invention are:

[0044] 1. By coordinating the structural deformation sensing unit, opening and closing status sensing unit, and environmental parameter sensing unit in the status sensing module, the synchronous acquisition of multi-dimensional status parameters such as door structural stress, opening and closing position, ambient temperature and humidity, and gas concentration is realized. This solves the technical problem of the existing technology having a single sensing dimension and being unable to fully reflect the comprehensive status of the air defense door, thereby significantly improving the completeness and coverage of status monitoring information.

[0045] 2. By using an edge computing module to perform local data preprocessing, spatiotemporal alignment, and feature extraction at the monitoring site, the raw data is transformed into standardized feature vectors before being transmitted remotely. This solves the technical problems of high data transmission latency and data loss due to unstable underground networks caused by cloud processing in existing technologies. Thus, while ensuring data processing accuracy, the status assessment response latency is reduced to the millisecond level, meeting the timeliness requirements of real-time monitoring of air defense doors.

[0046] 3. By using a multi-dimensional state assessment model trained on historical data in the state assessment module, the structural health, opening and closing function integrity, and environmental safety of the door are independently assessed and weighted and integrated. This solves the technical problem that fixed threshold judgments in the existing technology are difficult to identify progressive degradation and multi-factor coupled failures. As a result, quantitative assessment of the comprehensive state index of the air defense door is realized, and the accuracy of state degradation trend prediction and early fault warning is improved.

[0047] 4. By executing the feedback module, the early warning control command is directly driven to drive the actuator of the air defense door, and the execution result is fed back to the status perception module to form a closed loop. This solves the technical problems of the separation of monitoring and control links and the inability to automatically trigger protective actions in abnormal states in the existing technology. Thus, it realizes the automated closed-loop control from status monitoring to emergency execution, and improves the emergency response speed and reliability of the air defense door in the event of an emergency. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the overall architecture of the automatic monitoring system for air-raid shelter doors according to an embodiment of the present invention;

[0049] Figure 2 This is a flowchart of the automatic monitoring method for air defense doors according to an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] The system and method described in this embodiment aim to achieve real-time, comprehensive, automatic monitoring and intelligent control of the operating status of air defense doors. The specific implementation of the system is described first, followed by the specific implementation of the method, and finally, differentiated embodiments and comparative examples are given.

[0052] I. Detailed Implementation of the System

[0053] The overall architecture of the automatic monitoring system for air-raid shelter doors described in this embodiment is as follows: Figure 1 As shown, the system includes a state awareness module 100, an edge computing module 200, a state assessment module 300, an early warning decision module 400, and an execution feedback module 500. These modules are sequentially connected via wired or wireless communication links, forming a complete monitoring link from data acquisition to execution feedback.

[0054] 1. State Awareness Module 100

[0055] The status sensing module 100 serves as the system's data input, responsible for collecting multi-dimensional status data of the air defense door and its surrounding environment. This module includes a structural deformation sensing unit 110, an opening / closing status sensing unit 120, an environmental parameter sensing unit 130, and a data aggregation unit 140.

[0056] The structural deformation sensing unit 110 is a strain sensor array deployed at key load-bearing parts of the door. Specifically, it includes: resistive strain gauges deployed at the four corners and center of the main load-bearing frame of the door leaf to collect stress distribution data of the door structure; and laser displacement sensors deployed on the sealing surface between the door leaf and the door frame to collect deformation displacement data of the door. The input data source is the mechanical deformation of the door structure under its own weight, air pressure, and external forces. The core processing logic is to convert the resistance change of the strain gauges into stress values ​​and the laser ranging signal into displacement. The output data is the stress value sequence and displacement value sequence of each measuring point. This unit is connected to the data aggregation unit 140 via a shielded cable to transmit the collected data to the data aggregation unit 140 in real time.

[0057] The hardware configuration of the opening / closing state sensing unit 120 consists of an angle encoder and a displacement sensor installed at the door hinge shaft and locking mechanism. Specifically, it includes: an absolute rotary encoder installed at the upper and lower hinge shafts of the door to collect door opening angle data; and a linear displacement sensor installed at the latch of the locking mechanism to collect locking engagement depth data. The input data sources are the door opening action and the locking action; the core processing logic is to convert the encoder pulse signal into an angle value and the displacement sensor voltage signal into an engagement depth value; the output data are the real-time door opening angle and locking engagement depth; this unit is connected to the data aggregation unit 140 via an RS485 bus.

[0058] The environmental parameter sensing unit 130 consists of temperature and humidity sensors, gas concentration sensors, and vibration acceleration sensors deployed around the door frame. Specifically, it includes: a digital temperature and humidity sensor deployed above the door frame to collect ambient temperature and humidity data; an electrochemical carbon monoxide sensor and a catalytic combustion combustible gas sensor deployed on the sidewalls of the door frame to collect harmful gas concentration data; and a triaxial MEMS vibration acceleration sensor deployed on the back of the door to collect door vibration spectrum data. The input data sources are the environmental parameters of the underground space surrounding the door and the door's own vibration response. The core processing logic converts the analog signals from each sensor into digital values ​​via an ADC. The output data includes temperature, humidity, gas concentration, and a vibration acceleration time-domain sequence. This unit is connected to the data aggregation unit 140 via a CAN bus.

[0059] The data aggregation unit 140 is configured as an embedded data acquisition gateway in hardware and a real-time data acquisition program in software. Its core processing logic is as follows: it receives raw data from the structural deformation sensing unit 110, the opening / closing state sensing unit 120, and the environmental parameter sensing unit 130; it adds a unified timestamp based on the NTP protocol to each data set; it aggregates and packages the data according to a preset data frame format; and it transmits the packaged data stream to the edge computing module 200 via an Ethernet interface. This unit is connected to the edge computing module 200 via an industrial Ethernet communication connection.

[0060] 2. Edge computing module 200

[0061] The edge computing module 200 is configured as an edge computing gateway deployed near a civil defense door, with a built-in ARM architecture processor and local storage unit. Its software is an embedded data processing program. The module includes a data preprocessing unit 210, a spatiotemporal alignment unit 220, and a feature extraction unit 230, all connected sequentially.

[0062] The data preprocessing unit 210 receives the raw data stream from the data aggregation unit 140 and executes the following core processing logic: First, it uses a 3σ criterion-based outlier removal algorithm to mark and remove sampling points that exceed three times the standard deviation of the mean; second, it uses linear interpolation to imput the missing data points generated after removal; finally, it uses the Z-score standardization method to normalize the dimensions of the data, mapping data with different physical dimensions to a standard distribution with a mean of 0 and a standard deviation of 1. The input data is a timestamped multivariate raw data stream; the output data is a cleaned and standardized data sequence; this unit is connected to the spatiotemporal alignment unit 220 through a memory buffer.

[0063] The spatiotemporal alignment unit 220 receives a standardized data sequence from the data preprocessing unit 210 and executes the following core processing logic: First, it aligns the timestamps of each sensor data based on a unified time reference (using GPS / BeiDou timing or NTP network timing) to ensure that the multi-source data have the same time resolution (set to 1Hz in this embodiment); second, based on a pre-calibrated spatial coordinate mapping table, it maps the data of different measuring points on the gate to a unified three-dimensional spatial coordinate system to achieve spatial alignment. The input data is the cleaned standardized data sequence; the output data is a spatiotemporally aligned multidimensional data matrix; this unit is connected to the feature extraction unit 230 through a memory buffer.

[0064] The feature extraction unit 230 receives a multidimensional data matrix from the spatiotemporal alignment unit 220 and executes the following core processing logic: First, it extracts time-domain statistical features, including the mean, variance, kurtosis, skewness, and range of each dimension of the data; second, it performs a Fast Fourier Transform (FFT) on the vibration acceleration data to extract frequency-domain energy features, including the proportion of dominant frequency energy, the proportion of high-frequency energy, and the spectral entropy; finally, it uses wavelet packet decomposition to extract joint time-frequency features, including the energy distribution coefficients of each frequency band. The above three types of features are concatenated along fixed dimensions to generate a standardized feature vector sequence. The input data is the spatiotemporally aligned multidimensional data matrix; the output data is the standardized feature vector sequence; this unit is connected to the state assessment module 300 via an industrial Ethernet communication connection, transmitting the feature vector sequence to the state assessment module 300.

[0065] 3. Status Assessment Module 300

[0066] The hardware form of the status assessment module 300 is an assessment server deployed in the monitoring center, and the software form is a status assessment program. This module includes an assessment model storage unit 310, a comprehensive index calculation unit 320, and a report generation unit 330, which are connected in sequence.

[0067] The evaluation model storage unit 310 stores the multi-dimensional state evaluation model. This model includes three sub-models: a door structure health sub-model, employing a support vector regression (SVR) model trained based on historical strain and displacement data, with structural deformation-related features as input and a structural health score from 0 to 100 as output; an opening / closing functional integrity sub-model, employing a random forest model trained based on historical angle and meshing depth data, with opening / closing state-related features as input and a functional integrity score from 0 to 100 as output; and an environmental safety sub-model, employing a logistic regression model trained based on historical temperature, humidity, and gas data, with environmental parameter-related features as input and an environmental safety score from 0 to 100 as output. Each sub-model is trained offline based on no fewer than 5000 sets of historical monitoring samples and is incrementally updated periodically (quarterly) based on newly added monitoring data. This unit is connected to the comprehensive index calculation unit 320 via a local file system or database.

[0068] The comprehensive index calculation unit 320 receives a standardized feature vector sequence from the feature extraction unit 230 and executes the following core processing logic: First, the feature vector sequence is split into structural feature sub-vectors, opening / closing feature sub-vectors, and environmental feature sub-vectors according to feature categories; second, each sub-vector is input into its corresponding sub-model to obtain structural health score, functional integrity score, and environmental safety score; finally, based on preset weight coefficients (in this embodiment, structural health weight is set to 0.4, functional integrity weight to 0.3, and environmental safety weight to 0.3), a weighted fusion is performed to calculate the comprehensive status index of the air-raid shelter door. The input data is a standardized feature vector sequence; the output data is the comprehensive status index and scores for each dimension; this unit is connected to the report generation unit 330 through a memory buffer.

[0069] The report generation unit 330 receives the comprehensive status index and scores for each dimension from the comprehensive index calculation unit 320, and executes the following core processing logic: First, it classifies the status level according to the comprehensive status index value (90-100 is normal, 70-89 is at attention, 50-69 is warning, and 0-49 is dangerous), and generates a status level label; second, for dimensions below the threshold (set to 70 points in this embodiment), it extracts a list of abnormal parameters, listing specific abnormal measurement points and abnormal values; finally, it performs a 72-hour trend prediction of the comprehensive status index based on the time series prediction method (ARIMA model), generating trend prediction information. The above information is integrated into a structured status assessment report. The input data is the comprehensive status index and scores for each dimension; the output data is the status assessment report; this unit is connected to the early warning decision module 400 via a TCP / IP network, transmitting the report to the early warning decision module 400.

[0070] 4. Early Warning Decision Module 400

[0071] The early warning decision module 400 is implemented as a decision server deployed in the monitoring center in hardware form and as an early warning decision program in software form. This module receives a status assessment report from the report generation unit 330 and executes the following core processing logic: First, it parses the status level identifier and comprehensive status index in the report; second, it determines the early warning level according to preset multi-level threshold rules (normal level corresponds to no early warning, attention level corresponds to level 3 early warning, early warning level corresponds to level 2 early warning, and danger level corresponds to level 1 early warning); finally, it generates corresponding early warning control instructions based on the early warning level, including logging and notifying maintenance personnel for level 3 early warning, activating audible and visual alarms and restricting door opening for level 2 early warning, and automatically closing doors and activating emergency ventilation for level 1 early warning. The input data is the status assessment report; the output data is the early warning control instructions; this module is connected to the execution feedback module 500 via a control bus, transmitting instructions to the execution feedback module 500.

[0072] 5. Execution Feedback Module 500

[0073] The hardware of the execution feedback module 500 is a door execution controller and drive circuit, while the software is an execution control program. This module receives early warning control commands from the early warning decision module 400 and executes the following core processing logic: First, it parses the action type and action parameters in the command; second, it sends control signals to the electric locking device and hydraulic opening and closing mechanism of the air defense door via the drive circuit; finally, it collects the door status data after execution through the opening and closing status sensing unit 120 in the status sensing module 100, and feeds back the execution results, such as whether the locking is in place and whether the door is closed, to the status sensing module 100, forming a closed-loop monitoring link from sensing to execution. The input data is the early warning control command; the output data is the execution result feedback signal; this module is connected to the status sensing module 100 through a closed-loop feedback link.

[0074] II. Detailed Implementation of the Method

[0075] The overall process of the automatic monitoring method for air-raid shelter doors described in this embodiment is as follows: Figure 2 As shown, the system includes a state awareness step S1, an edge computing step S2, a state assessment step S3, an early warning decision step S4, and an execution feedback step S5. Each step is executed by the corresponding module of the system, and the steps and modules correspond one-to-one, all based on the same inventive concept.

[0076] S1: State-aware steps

[0077] Executed by the state-aware module 100, specifically including:

[0078] The structural deformation sensing unit 110 collects stress distribution data and deformation displacement data of the door structure at key load-bearing parts of the door body (the four corners and the middle of the main load-bearing frame of the door leaf and the sealing surface where the door leaf contacts the door frame), and the sampling frequency is set to 10Hz.

[0079] The opening and closing state sensing unit 120 collects door opening angle data and locking engagement depth data at the upper and lower hinge shafts of the door leaf and the lock tongue of the locking mechanism, with the sampling frequency set to 5Hz.

[0080] The environmental parameter sensing unit 130 collects ambient temperature data, ambient humidity data, harmful gas concentration data and door vibration spectrum data around the door frame. The sampling frequency for temperature, humidity and gas concentration is set to 1Hz, and the sampling frequency for vibration data is set to 100Hz.

[0081] The data aggregation unit 140 aggregates and packages the raw data collected by each sensing unit according to a unified timestamp (based on NTP protocol synchronization). The data frame format includes timestamp field, sensor ID field, data type field and data value field, and is transmitted to edge computing module 200 via industrial Ethernet.

[0082] S2: Edge computing steps

[0083] Executed by the edge computing module 200, specifically including:

[0084] The data preprocessing unit 210 performs outlier removal (3σ criterion), missing value imputation (linear interpolation), and dimensional normalization (Z-score standardization) on the received raw data stream.

[0085] The time-space alignment unit 220 aligns the timestamps of each sensor data based on a unified time reference (GPS / BeiDou time synchronization), with a unified time resolution of 1Hz, and maps the data of each measurement point to a unified three-dimensional spatial coordinate system based on a pre-calibrated spatial coordinate mapping table.

[0086] The feature extraction unit 230 extracts time-domain statistical features (mean, variance, kurtosis, skewness, range), frequency-domain energy features (primary frequency energy ratio, high frequency energy ratio, spectral entropy), and time-frequency joint features (wavelet packet decomposition of energy distribution coefficients of each frequency band) from the aligned data. The above features are spliced ​​together according to a fixed dimension to generate a standardized feature vector sequence with a vector dimension of 48, and transmitted to the status assessment module 300 via industrial Ethernet.

[0087] S3: Status Assessment Steps

[0088] Performed by the status assessment module 300, specifically including:

[0089] The evaluation model storage unit 310 provides a multi-dimensional state evaluation model, including a door structure health SVR sub-model, an opening and closing function integrity random forest sub-model, and an environmental safety logistic regression sub-model. Each sub-model is trained offline based on no less than 5,000 sets of historical samples.

[0090] The comprehensive index calculation unit 320 splits the 48-dimensional standardized feature vector sequence into structural feature sub-vectors (18-dimensional), open / closed feature sub-vectors (12-dimensional), and environmental feature sub-vectors (18-dimensional) according to feature categories. These sub-vectors are then input into the corresponding sub-models to obtain structural health scores, functional integrity scores, and environmental safety scores. The comprehensive state index is then calculated by weighted fusion based on weight coefficients (0.4, 0.3, 0.3).

[0091] The report generation unit 330 classifies the status level (90-100 normal, 70-89 attention, 50-69 warning, 0-49 danger) based on the comprehensive status index, extracts the list of abnormal parameters, and performs 72-hour trend prediction based on the ARIMA model to generate a structured status assessment report, which is then transmitted to the early warning decision module 400 via TCP / IP network.

[0092] S4: Early Warning Decision-Making Steps

[0093] The early warning decision module 400 executes the process, receiving the status assessment report, parsing the status level identifier and comprehensive status index, and generating an early warning level determination result based on multi-level threshold rules: Normal level generates no action command; Attention level generates log recording and notification command; Early warning level generates audible and visual alarm and door opening restriction command; Hazard level generates automatic door closing and emergency ventilation activation command. The early warning control command is then transmitted to the execution feedback module 500 via the control bus.

[0094] S5: Perform feedback steps

[0095] The execution feedback module 500 executes the command, receives the warning control command, analyzes the action type and parameters, sends the control signal to the electric locking device or hydraulic gate opener through the drive circuit, and collects the door opening angle and locking engagement depth data after execution through the opening and closing status sensing unit 120. The execution result is then fed back to the status sensing module 100 to form a monitoring closed loop. Specific Implementation

[0096] Example 1: Basic Standard Scenario (Entrance / Exit of Underground Civil Defense Project)

[0097] This embodiment is applied to a steel structure protective airtight door at the entrance of an underground civil defense project. The door has a size of 2.0m × 1.5m, a door leaf weight of about 800kg, and is equipped with an electric locking device.

[0098] Sensor deployment parameters: Five resistance strain gauges (model BX120-3AA, range ±3000με, accuracy ±1με) are installed at the four corners and center of the main load-bearing frame of the door leaf; two laser displacement sensors (model HL-G1-S500, range 500mm, accuracy ±0.1mm) are installed on the sealing surface between the door leaf and the door frame; one absolute rotary encoder (model EAM580-B, 16-bit resolution) is installed at each of the upper and lower hinge shafts; and one linear displacement sensor (model KTR-25m) is installed at the latch of the locking mechanism. m, measuring range 25mm, accuracy ±0.01mm; a digital temperature and humidity sensor (model SHT30, temperature accuracy ±0.3℃, humidity accuracy ±2%RH) is installed above the door frame; a carbon monoxide sensor (model MQ-7, measuring range 0-1000ppm) and a combustible gas sensor (model MQ-5, measuring range 0-10000ppm) are installed on the side wall of the door frame; a triaxial MEMS vibration acceleration sensor (model ADXL355, measuring range ±2g, resolution 20bit) is installed on the back of the door leaf.

[0099] System operating parameters: data acquisition gateway sampling period 100ms, edge computing gateway processor main frequency 1.2GHz, status assessment server configuration of 8-core CPU / 32GB memory / 1TB SSD, assessment model training sample size 6000 groups, weight coefficients set as structural health 0.4, functional integrity 0.3, and environmental safety 0.3.

[0100] Complete execution process: After the system is powered on, the state perception module 100 continuously collects multi-dimensional state data according to the above parameters. The data aggregation unit 140 generates a data frame every second and transmits it to the edge computing module 200. The edge computing module 200 performs data preprocessing, spatiotemporal alignment, and feature extraction locally, generating a 48-dimensional standardized feature vector every second and transmitting it to the state evaluation module 300. The state evaluation module 300 performs a comprehensive state index calculation every second and generates a state evaluation report. When the comprehensive state index drops below 70, the early warning decision module 400 generates a secondary early warning command, and the execution feedback module 500 drives the audible and visual alarm to start and restricts the electric door opening function; when the comprehensive state index drops below 50, a primary early warning command is generated, automatically triggering the electric locking device to close the door and start the emergency ventilation system.

[0101] Technical performance data: After 30 days of continuous operation, the average status assessment response latency was 85ms, data transmission bandwidth usage was reduced by 72% (compared to direct transmission of raw data), the average early warning time for progressive deformation of the door structure was 72 hours, and the false alarm rate was less than 3%.

[0102] Example 2: Differentiated Scenario (Civil Defense Partition Door in Subway Section)

[0103] This embodiment is applied to a civil defense isolation door in a subway tunnel. The door is 4.5m × 3.5m in size and weighs about 3500kg. It is equipped with a hydraulic opening and closing mechanism and is deployed in a semi-enclosed space inside the tunnel, where there is strong vibration interference when trains pass by.

[0104] The differences from Example 1 are as follows: the sampling frequency of the vibration acceleration sensor is increased to 500Hz to capture the broadband vibration characteristics when the train passes; two strain gauges are added to the connection between the door frame and the tunnel wall to monitor the anchorage status of the door frame; one PM2.5 sensor (model PMS5003) is added to the environmental parameter sensing unit to monitor the dust concentration in the tunnel; the edge computing module adds a vibration signal filtering preprocessing stage, using a bandpass filter (passband 5-200Hz) to remove low-frequency and high-frequency interference when the train passes; the weight coefficients of the evaluation model are adjusted to structural health 0.5, functional integrity 0.3, and environmental safety 0.2 to strengthen the weight of structural safety assessment.

[0105] Complete execution process: During train passage periods (detecting train approach characteristics via vibration acceleration sensors), the system temporarily increases the sampling frequency to 500Hz. The edge computing module 200 performs bandpass filtering to extract vibration frequency domain features and identify the dynamic response of the gate under train aerodynamic loads. When a shift in the gate's dominant vibration frequency is detected (the shift exceeds the baseline value by more than 15%) or an abnormal increase in the gate frame anchorage strain is detected (exceeding the baseline value by more than 20%), the condition assessment module 300 automatically increases the assessment weight of the structural health sub-model and generates a report of concern. During non-train passage periods, the system reverts to the normal 1Hz sampling frequency and executes the standard monitoring process.

[0106] Technical performance data: After 60 days of continuous operation, the system achieved a 96.5% accuracy rate in assessing the condition under train interference. It successfully identified early signs of loosening of the door frame anchor bolts (structural health score dropped from 92 to 78), with an early warning time of 96 hours, thus preventing further deterioration of the door under subsequent train vibration loads.

[0107] Comparison: Existing technical solutions

[0108] Using the same air-raid shelter door as in Example 1, a "Regional Air-raid Shelter Door Status Monitoring and Management System" based on existing technology is deployed. This system includes a sensor module (magnetic field sensor, temperature sensor, humidity sensor, carbon monoxide sensor, and triaxial sensor), a processor, a network module, and a host computer. Sensor data is processed simply by the processor and then transmitted to the host computer via the network module. The host computer determines the status based on fixed thresholds.

[0109] The comparison of the above technical effect data is shown in the table below:

[0110] Table 1

[0111]

[0112] The automatic monitoring system and method for air-raid shelter doors described in this invention can also be implemented as a computer-readable storage medium and an electronic device. Specifically, the computer program implementing the above method can be stored in a non-transitory computer-readable storage medium, such as ROM, RAM, flash memory, hard disk, optical disk, etc. When the program is executed by a processor, all steps of the above-mentioned automatic monitoring method for air-raid shelter doors are implemented. In addition, the functions of each module of the above system can be integrated into a single electronic device, such as an embedded industrial control computer, an industrial panel PC, or a dedicated monitoring terminal. The processor of this electronic device executes the corresponding program to achieve automatic monitoring and intelligent control of the status of air-raid shelter doors.

[0113] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automatic monitoring system for air-raid shelter doors, characterized in that, include: The state perception module (100) is used to collect multi-dimensional state data of the air defense door body and the surrounding environment. The multi-dimensional state data includes door structure deformation data, door opening and closing position data, ambient temperature and humidity data, ambient gas concentration data and door vibration spectrum data. The edge computing module (200) is communicatively connected to the state awareness module (100) and is used to receive the multi-dimensional state data, perform local data preprocessing, multi-source data spatiotemporal alignment and feature extraction, and generate a standardized feature vector sequence. The status assessment module (300) is communicatively connected to the edge computing module (200) and is used to receive the standardized feature vector sequence, calculate the comprehensive status index of the air defense door based on the preset multidimensional status assessment model, and generate a status assessment report. The early warning decision module (400) is communicatively connected to the status assessment module (300), and is used to receive the status assessment report, generate an early warning level determination result according to the preset multi-level threshold rules, and output the corresponding early warning control command; The execution feedback module (500) is communicatively connected to the early warning decision module (400) and is used to receive the early warning control command, drive the personnel defense door actuator to perform corresponding actions, and feed back the execution result to the status perception module (100) to form a monitoring closed loop.

2. The automatic monitoring system for air-raid shelter doors according to claim 1, characterized in that, The state-aware module (100) includes: The structural deformation sensing unit (110) is configured to deploy a strain sensor array at the key load-bearing parts of the door body to collect stress distribution data and deformation displacement data of the door body structure. The opening and closing state sensing unit (120) is configured to install an angle encoder and a displacement sensor at the door hinge shaft and locking mechanism to collect door opening angle data and locking engagement depth data. The environmental parameter sensing unit (130) is configured to install temperature and humidity sensors, gas concentration sensors and vibration acceleration sensors around the door frame to collect ambient temperature data, ambient humidity data, harmful gas concentration data and door vibration spectrum data. The data aggregation unit (140) is communicatively connected to the structural deformation sensing unit (110), the opening and closing state sensing unit (120), and the environmental parameter sensing unit (130), respectively, and is used to aggregate and package the raw data collected by each sensing unit according to a unified timestamp and transmit it to the edge computing module (200).

3. The automatic monitoring system for air-raid shelter doors according to claim 1, characterized in that, The edge computing module (200) includes: The data preprocessing unit (210) is used to perform outlier removal, missing value imputation and dimension normalization on the received raw data. The spatiotemporal alignment unit (220) is used to perform timestamp alignment of multi-source heterogeneous data based on a unified time reference, and to perform spatial position alignment of data from different measuring points of the door based on spatial coordinate mapping. The feature extraction unit (230) is used to extract time-domain statistical features, frequency-domain energy features, and time-frequency joint features from the aligned data to generate the standardized feature vector sequence. The data preprocessing unit (210), the spatiotemporal alignment unit (220), and the feature extraction unit (230) are connected in series, and the output of the feature extraction unit (230) is communicatively connected to the input of the state evaluation module (300).

4. The automatic monitoring system for air-raid shelter doors according to claim 1, characterized in that, The status assessment module (300) includes: The evaluation model storage unit (310) is used to store the multi-dimensional state evaluation model, which includes a door structure health sub-model, an opening and closing function integrity sub-model, and an environmental safety sub-model trained based on historical monitoring data. The comprehensive index calculation unit (320) is used to input the standardized feature vector sequence into each sub-model to obtain the status score of each dimension, and perform weighted fusion based on the preset weight coefficient to calculate the comprehensive status index of the civil defense door; The report generation unit (330) is used to generate the status assessment report containing status level identifiers, abnormal parameter lists and trend prediction information based on the comprehensive status index of the air defense door and the status scores of each dimension. The evaluation model storage unit (310), the comprehensive index calculation unit (320), and the report generation unit (330) are connected in sequence, and the output end of the report generation unit (330) is communicatively connected to the input end of the early warning decision module (400).

5. An automatic monitoring method for air-raid shelter doors, characterized in that, Includes the following steps: S1: State perception step, where the state perception module (100) collects multi-dimensional state data of the human defense door body and the surrounding environment. The multi-dimensional state data includes door structure deformation data, door opening and closing position data, ambient temperature and humidity data, ambient gas concentration data and door vibration spectrum data. S2: Edge computing step, where the edge computing module (200) receives the multi-source state data, performs local data preprocessing, multi-source data spatiotemporal alignment and feature extraction, and generates a standardized feature vector sequence; S3: Status assessment step, where the status assessment module (300) receives the standardized feature vector sequence, calculates the comprehensive status index of the air defense door based on the preset multi-dimensional status assessment model, and generates a status assessment report; S4: Early warning decision-making step, whereby the early warning decision-making module (400) receives the status assessment report, generates an early warning level determination result according to the preset multi-level threshold rules, and outputs the corresponding early warning control command; S5: Execution feedback step, the execution feedback module (500) receives the early warning control command, drives the personnel defense door actuator to perform the corresponding action, and feeds back the execution result to the status perception module (100) to form a monitoring closed loop.

6. The automatic monitoring method for air-raid shelter doors according to claim 5, characterized in that, The S1 state awareness step includes: The structural deformation sensing unit (110) collects stress distribution data and deformation displacement data of the gate structure at key load-bearing parts of the gate; The door opening angle data and locking engagement depth data are collected by the opening and closing state sensing unit (120) at the door hinge shaft and locking mechanism; The environmental parameter sensing unit (130) collects ambient temperature data, ambient humidity data, harmful gas concentration data and door vibration spectrum data around the door frame; The data aggregation unit (140) aggregates and packages the raw data collected by each sensing unit according to a unified timestamp and transmits it to the edge computing module (200).

7. The automatic monitoring method for air-raid shelter doors according to claim 6, characterized in that, The S2 edge computing steps include: The data preprocessing unit (210) performs outlier removal, missing value imputation and dimension normalization on the received raw data; The spatiotemporal alignment unit (220) performs timestamp alignment on multi-source heterogeneous data based on a unified time reference, and performs spatial position alignment on data of different measuring points of the door based on spatial coordinate mapping; The feature extraction unit (230) extracts time-domain statistical features, frequency-domain energy features, and time-frequency joint features from the aligned data to generate the standardized feature vector sequence.

8. The automatic monitoring method for air-raid shelter doors according to claim 7, characterized in that, The S3 state assessment steps include: The multidimensional state evaluation model is provided by the evaluation model storage unit (310), including the door structure health sub-model, the opening and closing function integrity sub-model and the environmental safety sub-model trained based on historical monitoring data; The standardized feature vector sequence is input into each sub-model by the comprehensive index calculation unit (320) to obtain the status scores of each dimension, and the comprehensive status index of the civil defense door is calculated by weighted fusion based on the preset weight coefficient. The report generation unit (330) generates a status assessment report containing status level identifiers, a list of abnormal parameters, and trend prediction information based on the comprehensive status index of the air defense door and the status scores of each dimension.

9. The automatic monitoring system for air-raid shelter doors according to any one of claims 1 to 4, characterized in that, The system is applied to real-time status monitoring scenarios of entrances and exits of underground civil defense projects, civil defense partition doors in subway sections, or protective doors of important infrastructure. The deployment environment is underground enclosed space or semi-enclosed passage, and the sensor placement position of the status perception module (100) is configured differently according to the mechanical characteristics of the door structure and the distribution characteristics of the ambient airflow.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the automatic monitoring method for air defense doors as described in any one of claims 5 to 8.

Images