Remote duty monitoring method and system

Abstract

The application provides a remote duty monitoring method and system, the method comprising: according to the environmental interaction parameter, the geometric analysis is carried out to the angle between the door body orientation angle and the airflow velocity vector, the potential smoke diffusion risk level is determined in combination with the gap width interval; according to the judgment result of the high-risk combination, the feature data of the similar scene is extracted from the pre-established historical case matching database, the corresponding risk scene label is generated, and the disposal priority is determined; according to the judgment result of the accelerated smoke diffusion channel, the airflow interaction intensity and the smoke concentration reading are fused to determine the high-risk scene judgment result; if it is judged as a high-risk scene, the disposal protocol sequence corresponding to the disposal priority and the risk scene label is retrieved from the database, real-time response instructions are generated, the fire door state is automatically adjusted, and the optimized safety configuration state is obtained.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a remote duty monitoring method and system. Background Technology

[0002] Fire doors are critical facilities for building fire prevention and control, and their smoke and fire isolation effects are directly related to the safety of personnel evacuation. In real-world scenarios, when a fire door is on the windward side and has tiny gaps, airflow can accelerate the intrusion of smoke or flames through the narrow gap effect. This combination poses a risk that is even greater than that of a large gap, but traditional remote monitoring methods are difficult to effectively address this core issue.

[0003] First, traditional monitoring cannot accurately identify the high-risk combination of "windward side + tiny gap". The orientation of the door needs to be collected by sensors to collect angle data, but the airflow direction is frequently affected by the ambient wind and the building's ventilation system. This dynamic fluctuation can interfere with the judgment of the windward side. At the same time, the tiny gaps in fire doors are difficult to detect by conventional monitoring methods due to their small size. Especially in complex lighting or obstructed environments, this risk feature may be completely missed, resulting in high-risk scenarios not being identified in time, and subsequent handling is out of the question.

[0004] Secondly, even if some risk characteristics are barely detected, there is a lack of dynamic tracking and comprehensive judgment mechanisms for risk combinations. Fire-related data such as airflow intensity, smoke concentration, and temperature are all in dynamic flux. Traditional methods cannot continuously track the fluctuation trends of these data, nor can they integrate multi-dimensional data to determine whether the risk is escalating. This leads to an inability to accurately match the corresponding handling priorities and response instructions, either triggering ineffective operations or delaying critical emergency adjustments, ultimately failing to achieve effective management of high-risk scenarios.

[0005] In summary, the failure to identify high-risk combinations is a fundamental obstacle, while the lack of dynamic tracking and comprehensive judgment prevents the effective handling of identified risks. There is an urgent need for a coherent technical solution that comprehensively addresses both "identification" and "judgment" to ensure that high-risk scenarios can be accurately captured and responded to in a timely manner. Summary of the Invention

[0006] To address the shortcomings and defects of existing technologies, this invention provides a remote duty monitoring method, mainly comprising:

[0007] By deploying a multi-point sensor array around the fire door, data such as the door's orientation angle, airflow velocity vector, airflow interaction intensity, gap width range, temperature distribution, and smoke concentration are collected in real time. The raw data is then denoised to obtain environmental interaction parameters.

[0008] Geometric analysis is performed on the angle between the door orientation angle and the airflow velocity vector based on environmental interaction parameters, and the potential smoke diffusion risk level is determined by combining the gap width range.

[0009] By combining the potential flue gas diffusion risk level with the dynamic change frequency and temperature distribution, the fluctuation of environmental interaction parameters is tracked and analyzed using the time series comparison method to determine whether there is a high-risk combination of windward risk marker and small gap characteristic value superposition.

[0010] Based on the high-risk combination judgment results, similar scenario feature data are extracted from the historical case matching database to generate risk scenario labels and determine priorities;

[0011] Based on the high-risk combination judgment results, real-time image data from the remote monitoring platform is obtained. Boundary extraction analysis is performed on the door gap area to identify the subtle change trend of the gap width and determine whether there is an accelerated flue gas diffusion channel.

[0012] The high-risk scenario assessment result is determined by combining the results of the accelerated flue gas diffusion channel assessment with the airflow interaction intensity and smoke concentration readings.

[0013] When a high-risk scenario is identified, a real-time response instruction is generated by retrieving the handling protocol sequence corresponding to the priority and risk scenario label from the database, and the fire door status is automatically adjusted to obtain an optimized safety configuration status.

[0014] Based on the optimized safety configuration status, the dynamic change frequency records in the monitoring log are updated and the adjusted environmental interaction parameters are transmitted to the central processing unit. The attenuation trend of airflow interaction intensity is verified cyclically to determine whether the overall risk has been reduced to a controllable range.

[0015] Furthermore, the environmental interaction parameters are obtained by real-time data collection of door orientation angle, airflow velocity vector, airflow interaction intensity, gap width range, temperature distribution, and smoke concentration readings through a multi-point sensor array deployed around the fire door. The raw data is then denoised to obtain the environmental interaction parameters, including:

[0016] A sensor array is arranged in a regular hexagonal grid around the fire door. The sensor network topology and spatial coordinates are obtained by triangulation based on the known distance and angle sensor readings between adjacent nodes.

[0017] Simultaneously collect tilt angle data, airflow velocity vector, airflow interaction intensity, temperature distribution data, gap width range data, and smoke concentration readings. Perform coordinate transformation on the airflow velocity vector to calculate the cosine value of the angle between the door normal vector and the airflow velocity vector and determine whether it is a windward state.

[0018] If the condition is determined to be windward, a Kalman filter is used to suppress noise in various data, and the filtered data is fused by a weighted average method to obtain the environmental interaction parameters.

[0019] Furthermore, the geometric analysis of the angle between the door orientation angle and the airflow velocity vector based on environmental interaction parameters, combined with the gap width range, determines the potential smoke diffusion risk level, including:

[0020] Extract the door orientation angle and airflow velocity vector from the environmental interaction parameters, obtain the cosine value of the included angle through vector dot product operation, and calculate the actual included angle value using the inverse cosine function;

[0021] The corresponding permeability coefficient value is retrieved from a pre-established flue gas permeability coefficient database, and the potential flue gas diffusion risk level is determined by comparing the flue gas permeability coefficient with a preset risk threshold set.

[0022] Furthermore, by combining the potential flue gas diffusion risk level with the dynamic change frequency and temperature distribution, and using a time series comparison method to track and analyze the fluctuations of environmental interaction parameters, it is determined whether there is a high-risk combination of windward risk markers and small gap characteristic values ​​superimposed, including:

[0023] Environmental interaction parameters are continuously collected based on the potential flue gas diffusion risk level, and a time series array is constructed.

[0024] The dynamic change frequency is obtained by using the sliding time window method to count the ratio of the number of changes of each parameter to the length of the time window. When the angle between the door angle and the airflow velocity vector is less than the preset value, a risk mark is generated on the windward side. When the gap width is in a small range, a small gap feature value is generated.

[0025] The change in temperature distribution is obtained by differential operation, the rate of change in temperature distribution is calculated, and the environmental parameter fluctuation index is obtained by weighted summation of dynamic change frequency and temperature distribution change rate and normalization. The fluctuation anomaly degree is obtained by comparing with historical data series of the same period.

[0026] When the windward risk marker and the small gap characteristic value coexist, and the environmental parameter fluctuation index and fluctuation anomaly degree both exceed the corresponding thresholds, a high-risk combination is determined to exist.

[0027] Furthermore, the step of extracting similar scenario feature data from the historical case matching database based on the high-risk combination judgment result to generate risk scenario labels and determine priorities includes:

[0028] Extract the feature vector of the current environment parameters, and calculate the similarity value with the feature vector of historical cases using the cosine similarity algorithm;

[0029] Read historical case records of fire development stages, smoke diffusion rates, and door damage levels that exceed the similarity threshold, and generate risk scenario labels;

[0030] By querying the preset hazard level table and response timetable using risk scenario tags, a comprehensive score is calculated to determine the priority of handling.

[0031] Furthermore, the step of obtaining real-time image data from the remote monitoring platform based on the high-risk combination judgment result, performing boundary extraction analysis on the door gap area to identify subtle changes in gap width, and determining whether there is an accelerated smoke diffusion channel includes:

[0032] Continuous image frames are extracted from the real-time video stream of the fire door area obtained from the remote monitoring platform, and then grayscale and histogram equalization are performed to obtain the preprocessed image;

[0033] The region of interest for the door gap is determined by detecting the straight line features of the door frame using Hough transform, and the door gap contour is extracted by Canny edge detection within the region of interest to obtain the set of boundary pixels.

[0034] Calculate the pixel distance between the left and right boundaries and convert it into a physical width value. Construct a time series width array in chronological order. Use a sliding window to traverse and calculate the rate of change of the width value within the window. When the rate of change of the width shows an increasing trend in multiple consecutive windows, it is marked as an accelerated flue gas diffusion channel.

[0035] Furthermore, the step of determining the high-risk scenario assessment result by integrating the airflow interaction intensity and smoke concentration readings based on the accelerated smoke diffusion channel assessment result includes:

[0036] The data obtained from the sensor network, including airflow interaction intensity and smoke concentration readings, are combined to form real-time monitoring data pairs, which are then matched with a pre-established two-dimensional risk assessment table.

[0037] When the airflow interaction intensity exceeds a preset intensity threshold and the smoke concentration exceeds a preset concentration threshold, it is determined as a high-risk scenario.

[0038] Furthermore, when a high-risk scenario is identified, the step of retrieving the handling protocol sequence corresponding to the priority and risk scenario label from the database to generate a real-time response instruction, and automatically adjusting the fire door status to obtain an optimized safety configuration status, includes:

[0039] If a scenario is determined to be high-risk, a query index is built based on priority and risk scenario tags to retrieve the protocol sequence from the handling protocol database.

[0040] The parameter values ​​in the protocol sequence are converted to form a real-time response instruction set.

[0041] The real-time response instruction set is sent to the execution unit through a preset communication interface, the return signal is monitored, and the optimized security configuration status is confirmed.

[0042] Furthermore, the step of updating the dynamic change frequency record in the monitoring log according to the optimized security configuration status and transmitting the adjusted environmental interaction parameters to the central processing unit, and cyclically verifying the attenuation trend of airflow interaction intensity, includes:

[0043] Based on the optimized safety configuration status, the current door angle and sealing pressure are written into the monitoring log, and the time change rate is calculated to obtain the dynamic change frequency record.

[0044] The dynamically changing frequency records and the re-acquired and adjusted environmental interaction parameters are packaged and transmitted to the central processing unit. The airflow interaction intensity values ​​are read cyclically to calculate the difference. When the intensity value continuously decreases and drops below the safety threshold, it is determined that the overall risk has been reduced to a controllable range.

[0045] A remote duty monitoring system, the system comprising:

[0046] The data acquisition and preprocessing module is used to collect preliminary environmental interaction parameters in real time through a multi-point sensor array around the fire door, and generate environmental interaction parameters after noise reduction.

[0047] The risk level analysis module is used to analyze the angle between the door and the airflow and the gap width based on environmental interaction parameters to determine the risk level of flue gas diffusion.

[0048] The high-risk combination judgment module is used to combine risk level, dynamic change frequency and temperature data, track environmental parameter fluctuations through time series comparison, and determine whether there is a high-risk combination of windward surface and small gap;

[0049] The priority determination module is used to extract similar scenario features from the historical case database based on high-risk combination results, generate risk scenario labels, and determine the priority of disposal.

[0050] The image analysis module is used to acquire real-time images from remote monitoring based on high-risk combination results, perform boundary extraction analysis on the door gap area, identify subtle changes in gap width, and determine whether an accelerated flue gas diffusion channel exists.

[0051] The high-risk scenario determination module is used to determine high-risk scenarios by combining the results of the accelerated smoke diffusion channel judgment with the airflow intensity and smoke concentration.

[0052] The response execution module is used in high-risk scenarios to retrieve the handling protocols corresponding to the priority and risk scenario tags, generate real-time instructions, and automatically adjust the fire door to a safe configuration state.

[0053] The status update and verification module is used to update the dynamic change frequency of the monitoring log according to the security configuration status, transmit the adjusted environmental interaction parameters, cyclically verify the airflow intensity attenuation trend, and determine whether the overall risk is controllable.

[0054] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0055] This invention discloses a remote monitoring method and system. By collecting multi-dimensional data such as door orientation, airflow, temperature, and smoke in real time, and integrating geometric analysis and time series comparison, it accurately identifies high-risk combinations of windward risks and small gap features, solving the operational challenge of dynamically assessing and responding promptly to smoke diffusion risks from fire doors in complex environments. This invention captures minute changes in door gaps using image boundary extraction technology, combines airflow intensity and smoke concentration data to determine accelerated diffusion channels, generates high-risk scenario labels, and automatically adjusts the fire door status by matching handling protocols from a historical case database, achieving optimized safety configuration. Simultaneously, this invention cyclically verifies risk attenuation trends to ensure overall risk is reduced to a controllable range. Its core technological effect lies in significantly improving the dynamic control capability of fire doors against smoke diffusion through multi-source data fusion and intelligent analysis, ensuring the safety of personnel and property. Attached Figure Description

[0056] Figure 1 This is a flowchart of a remote duty monitoring method according to an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of the structure of a remote duty monitoring system according to an embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram of the high-risk combination judgment logic of a remote duty monitoring method according to an embodiment of this application;

[0059] Figure 4 This is a schematic diagram illustrating the risk verification attenuation trend of a remote duty monitoring method according to an embodiment of this application. Detailed Implementation

[0060] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0061] This application proposes a remote monitoring method that addresses the core issues of traditional fire door monitoring methods, such as the inability to accurately identify the high-risk combination of "windward side and micro-gap superposition" and the lack of tracking and intelligent response capabilities for dynamic risk evolution. By deploying a sensor array to acquire multi-dimensional data in real time, including door orientation, airflow, gaps, temperature, and smoke concentration, the method first integrates geometric analysis and time series comparison to dynamically identify and track the high-risk combination formed by the superposition of "windward side risk markers" and "micro-gap feature values." Then, based on the high-risk combination, a dual verification is triggered: on the one hand, image analysis accurately captures the subtle changing trends of the door gaps to confirm accelerated diffusion channels; on the other hand, similar scenarios are matched from a historical case database to generate risk labels and handling priorities. Subsequently, high-risk scenarios are determined by comprehensively considering airflow intensity and smoke concentration data, and matching handling protocols are automatically executed according to the aforementioned labels and priorities, adjusting the fire door status in real time. Finally, a closed-loop control is formed to continuously verify the effectiveness of protective measures until the risk is under control. It has achieved a fundamental transformation in the risk of smoke spread from fire doors, from passive monitoring to proactive early warning, from static judgment to dynamic tracking, and from manual handling to intelligent response. It has significantly improved the accuracy and timeliness of blocking the spread of smoke in complex fire environments, thereby effectively ensuring the safety of personnel evacuation and the preservation of property.

[0062] For example, this embodiment provides a remote duty monitoring method, which may include the following steps not shown in the figures:

[0063] A01. Obtain raw data from a multi-point sensor array deployed around the fire door, and obtain environmental interaction parameters by denoising the raw data. Environmental interaction parameters include: door orientation angle, airflow velocity vector, airflow interaction intensity, gap width range data, temperature distribution, and smoke concentration reading.

[0064] For example, a sensor array is arranged around a fire door in a regular hexagonal grid. The sensor network topology and spatial coordinates are obtained by triangulation using known distances and angle sensor readings between adjacent nodes. Simultaneously, tilt angle data, airflow velocity vector, airflow interaction intensity, temperature distribution data, gap width range data, and smoke concentration readings are collected. The airflow velocity vector is transformed by coordinates to calculate the cosine of the angle between the door normal vector and the airflow velocity vector, and it is determined whether the door is facing the wind. If it is determined to be facing the wind, a Kalman filter is used to suppress noise in each data point. The filtered data is then fused using a weighted average method to obtain the environmental interaction parameters.

[0065] A02. Determine the potential flue gas diffusion risk level based on environmental interaction parameters;

[0066] A03. Based on the potential flue gas diffusion risk level, combined with the dynamic change frequency and temperature distribution, a high-risk combination judgment result is generated by superimposing the windward risk mark and the small gap characteristic value.

[0067] A04. Based on the similar scene feature data extracted from the high-risk combination judgment results, generate risk scene labels and determine priorities; and based on the real-time images obtained from the remote monitoring platform and the high-risk combination judgment results, determine whether there is an accelerated flue gas diffusion channel.

[0068] A05. Based on the judgment results of the accelerated smoke diffusion channel, the high-risk scenario judgment results are determined by integrating the airflow interaction intensity and smoke concentration readings; when a high-risk scenario is determined, the status of the fire door is adjusted based on the high-risk judgment results to obtain the optimized safety configuration status and the adjusted environmental interaction parameters.

[0069] A06. Based on the optimized safety configuration status and the adjusted environmental interaction parameters, cyclically verify the attenuation trend of airflow interaction intensity to determine whether the overall risk has been reduced to a controllable range.

[0070] This embodiment achieves, for the first time, accurate identification of the dynamic high-risk combination of "windward side" and "minor gaps" by simultaneously collecting and fusing multi-dimensional data such as door orientation, airflow, gaps, temperature, and smoke concentration, combined with time series analysis. Furthermore, it verifies the diffusion channel through image analysis and generates intelligent response strategies based on historical case matching, realizing integrated decision-making from risk perception and dual verification to precise response. Finally, through a closed-loop verification mechanism, it continuously tracks risk attenuation, upgrading fire doors from passive isolation facilities to active safety systems with dynamic early warning, intelligent response, and adaptive control capabilities. This fundamentally solves the problems of missed, misjudged, and delayed responses to specific high-risk situations in traditional methods, significantly improving the accuracy, timeliness, and reliability of fire smoke diffusion control.

[0071] In addition, such as Figure 1 As shown in this embodiment, a remote duty monitoring method may specifically include:

[0072] S101. By deploying a multi-point sensor array around the fire door, real-time data such as the door's orientation angle, airflow velocity vector, airflow interaction intensity, gap width range, temperature distribution, and smoke concentration are collected. The collected raw data is then denoised to obtain environmental interaction parameters.

[0073] A sensor array is arranged in a regular hexagonal grid around the fire door. Each node is equipped with an angle sensor, an anemometer, an infrared thermometer, a laser rangefinder, and a smoke concentration sensor. Triangulation calculations are performed using known distances between adjacent nodes and angle sensor readings to obtain the sensor network topology and the spatial coordinates of each node. Based on the sensor network topology and spatial coordinates, the tilt angle data from the door angle sensor, the airflow velocity vector from the anemometer, the temperature distribution data from the infrared thermometer, the width measurement of the door gap from the laser rangefinder, and the smoke concentration reading from the smoke concentration sensor are collected synchronously. Coordinate transformation is performed on the airflow velocity vector, and the cosine value of the angle between the door normal vector and the airflow velocity vector is calculated. When the cosine value of the angle is greater than a preset threshold, it is determined to be a windward state. If the condition is determined to be windward, a Kalman filter is used to suppress noise in the tilt angle data, airflow velocity vector, temperature distribution data, gap width interval data, and smoke concentration readings. The filtered data are then fused using a weighted average method. The magnitude of the airflow velocity vector is used as the airflow interaction intensity, the gap width interval data is divided into preset width intervals, and the temperature distribution data is converted into a temperature distribution field to obtain the environmental interaction parameters.

[0074] Specifically, in one implementation, the hexagonal grid arrangement adopts a honeycomb structure, with the spacing between adjacent nodes set at 2 meters. Each node is installed on the perimeter wall of the fire door frame, forming a sensing coverage area surrounding the door. Triangulation calculation is based on the known node spacing and the relative azimuth angle measured by the angle sensor. The precise position of each node in a three-dimensional coordinate system with the center of the door as the origin is determined by the law of cosines.

[0075] Specifically, the coordinate transformation of the airflow velocity vector employs a rotation matrix method, converting the velocity components in the local coordinate system measured by the anemometer to the door coordinate system. The door normal vector is determined by the pitch and yaw angles measured by the angle sensor, and is then multiplied by the transformed airflow velocity vector to obtain the cosine of the included angle. When this value is greater than 0.7, the door is determined to be in a windward state, at which point the risk of flue gas infiltration through the door gaps increases significantly.

[0076] It should be noted that the Kalman filter establishes a state-space model during noise suppression, using sensor measurements as observation vectors. Through two recursive steps of prediction and update, it eliminates random noise caused by environmental vibrations, electromagnetic interference, and other factors. The filter's state transition matrix is ​​adjusted according to the sensor sampling frequency and data variation characteristics, while the observation noise covariance matrix is ​​predetermined through sensor calibration experiments. For the gap width range data, a laser rangefinder scans along the vertical direction of the door gap to acquire multi-point distance data. The door gap contour is fitted using the least squares method, dividing the width values ​​into preset ranges such as 0-5 mm, 5-10 mm, and 10-15 mm, each corresponding to a different flue gas permeability coefficient. Temperature distribution data is obtained from the door surface temperature field using an infrared thermometer array. The partial derivatives of the temperature at each point with respect to spatial coordinates are calculated using the finite difference method to construct a temperature distribution field. The distribution direction indicates the heat flow transfer path, and the distribution magnitude reflects the heat transfer intensity.

[0077] The above-mentioned optimized sensor layout, precise spatial positioning, advanced noise reduction algorithms and multi-source data fusion technology have enabled high-precision, high-reliability, and all-round real-time perception of the environmental status around fire doors, laying a solid data foundation for subsequent risk identification, scene determination and intelligent response, and improving the overall monitoring accuracy and decision reliability of the system from the source.

[0078] Preferably, the weighting coefficients of each data item in the weighted average fusion are dynamically adjusted according to the sensor accuracy level and data stability. The airflow interaction intensity is directly taken from the Euclidean norm of the airflow velocity vector, reflecting the degree of impact of the airflow on the door. The environmental interaction parameters include five components: windward surface identification, gap level, temperature distribution vector, airflow intensity value, and smoke concentration value, which constitute a feature vector describing the environmental state around the fire door.

[0079] This design ensures that in complex and ever-changing real-world environments, the system can automatically assign greater decision-making influence to more reliable and accurate sensor data, effectively suppressing overall misjudgments caused by accidental errors or performance fluctuations of individual sensors, and improving system robustness. Simultaneously, the processed multi-source data is encapsulated into a standardized feature vector containing five dimensions: windward surface identification, gap level, temperature distribution vector, airflow intensity value, and smoke concentration value. Essentially, this transforms the continuous and chaotic state of the physical world into discrete, structured, machine-readable semantics. Achieving a precise digital representation of the fuzzy concept of "environmental state," enabling logical reasoning and intelligent matching based on explicit features, is a key technological step in realizing the leap from "data acquisition" to "intelligent cognition."

[0080] S102. Based on the environmental interaction parameters, perform geometric analysis on the angle between the door orientation angle and the airflow velocity vector, and determine the potential smoke diffusion risk level by combining the gap width range.

[0081] The door orientation angle and airflow velocity vector are extracted from environmental interaction parameters. The cosine value of the angle between the two vectors is obtained through vector dot product operation. The actual angle value is calculated using the inverse cosine function. When the angle value is less than a preset angle threshold, it is determined to be in a positive alignment state, thus obtaining the result of the degree of alignment between the airflow and the door surface. Based on the combination relationship between the degree of alignment determination result and the gap width range, the corresponding permeability coefficient value is retrieved from a pre-established flue gas permeability coefficient database. The database is indexed according to the angle range and gap width range, and each index position stores the flue gas permeability coefficient measured experimentally. The flue gas permeability coefficient is compared with a preset risk threshold set, which includes three values: high risk threshold, medium risk threshold, and low risk threshold. When the permeability coefficient is greater than the high risk threshold, it is marked as high-risk; when it is between the medium and high risk thresholds, it is marked as medium-risk; and when it is less than the medium risk threshold, it is marked as low-risk, thus determining the potential flue gas diffusion risk level.

[0082] This embodiment combines the spatial geometric relationship between the door and airflow with the physical dimensions of the door gap, and uses a smoke permeability coefficient database based on experimental data for matching and querying. This transforms the assessment of smoke diffusion risk from qualitative, empirical judgment to quantitative, scientific evaluation. The method transforms the ambiguous state of "whether the airflow is directly facing the door" into a calculable angle value, which is then linked to the key risk parameter of the gap width. A lookup table is used to quickly derive a coefficient value reflecting the actual permeability, which is finally compared with a clearly defined risk threshold to output a high, medium, or low risk level. This process not only improves the objectivity, consistency, and repeatability of risk assessment but also provides accurate and reliable decision-making basis for subsequent risk-based response, enabling the identification of potential smoke diffusion threats in the early stages of a fire and buying crucial time for targeted prevention and control measures.

[0083] Specifically, in one embodiment, the door orientation angle is obtained using a tilt sensor mounted on the door frame, which outputs the deflection angle of the door normal relative to the vertical direction. The airflow velocity vector is constructed from the three-axis component measurements of an anemometer, with each component representing the velocity projection of the airflow on the corresponding coordinate axis.

[0084] Specifically, the vector dot product operation is performed according to the vector dot product formula. The three components of the door's normal vector are multiplied by the corresponding components of the airflow velocity vector, and then summed. The resulting scalar value is divided by the product of the magnitudes of the two vectors to obtain the cosine of the included angle. The cosine value is then converted to an angle value using the inverse cosine function. When this angle is less than 30 degrees, the airflow is considered to be almost perpendicular to the door surface, indicating a perfectly aligned state; an angle between 30 and 60 degrees indicates a moderately aligned state; and an angle exceeding 60 degrees indicates a non-aligned state.

[0085] It should be noted that the smoke permeability coefficient database is based on experimental data from smoke infiltration through fire door gaps. Different combinations of door gap widths and airflow angles were used in the experiment. The amount of smoke passing through the door gap per unit time was measured using a smoke generator and a concentration detector, and the permeability coefficient was calculated. The database is stored using a two-dimensional matrix structure. Row indices correspond to the angle range, column indices correspond to the gap width range, and matrix elements are the permeability coefficient values ​​under the corresponding conditions. During a query, the database locates the corresponding matrix position based on the currently measured angle and gap width values ​​and directly retrieves the permeability coefficient.

[0086] Preferably, the risk threshold set is set with reference to building fire protection codes and historical fire data statistics. The high-risk threshold corresponds to a smoke penetration rate exceeding 80% of the rated smoke isolation capacity of the fire door per minute, the medium-risk threshold is 50%, and the low-risk threshold is 20%. When the detected penetration coefficient is multiplied by the current smoke concentration and airflow intensity, the estimated actual penetration rate is obtained, and the risk level is determined by comparing it with the threshold.

[0087] In one possible implementation, for high-risk situations, the automatic inflation mechanism of the fire door's sealing strip is triggered to enhance the door's sealing performance; for medium-risk situations, local smoke extraction facilities are activated; and for low-risk situations, only monitoring logs are recorded. This tiered response mechanism enables adaptive protection of fire doors under different fire scenarios.

[0088] This embodiment uses tilt sensors and an anemometers to acquire the door's orientation and airflow vector, and then calculates the angle between them through vector operations, thus objectively quantifying the degree of "windwardness." Combined with the measured door gap width, a two-dimensional database based on smoke infiltration experiments is consulted to quickly obtain the corresponding smoke permeability coefficient. Finally, this coefficient is compared with the risk thresholds set in the standards, automatically outputting a high, medium, or low risk level. The entire process achieves quantitative and coherent assessment from multi-source data acquisition to risk level output, laying the foundation for accurate response.

[0089] S103. By combining the potential flue gas diffusion risk level with the dynamic change frequency and temperature distribution, the fluctuation of environmental interaction parameters is tracked and analyzed using the time series comparison method to determine whether there is a high-risk combination of windward risk marker and small gap characteristic value superposition.

[0090] Based on the potential flue gas diffusion risk level, environmental interaction parameters are continuously collected at preset sampling intervals to construct a time series array containing risk level, door angle, airflow intensity, gap width, and temperature data, resulting in the original monitoring sequence. For the original monitoring sequence, a sliding time window method is used to segment the sequence. The ratio of the number of changes in each parameter's value within adjacent time windows to the window length is used as the dynamic change frequency. When the angle between the door angle and the airflow velocity vector is less than a preset threshold, a windward risk marker is generated; when the gap width is within a preset small interval, a small gap feature value is generated. The change in temperature distribution between adjacent time points in the time series array is obtained through differential operations. The rate of change in temperature distribution is calculated, and the dynamic change frequency and the rate of change in temperature distribution are weighted, summed, and normalized to obtain the environmental parameter fluctuation index. This index is then compared with historical data from the same period using Pearson correlation to obtain the fluctuation anomaly degree. If both the windward risk marker and the small gap feature value exist simultaneously, and the environmental parameter fluctuation index exceeds a preset high-risk threshold, while the fluctuation anomaly degree exceeds a preset anomaly threshold, then a high-risk combination is identified.

[0091] This embodiment continuously calculates the dynamic change frequency of each parameter through a sliding window and integrates the temperature change rate to generate a "fluctuation index" reflecting the degree of environmental instability. Simultaneously, the system not only checks whether the "windward side" and "small gap" occur simultaneously, but also requires that both the fluctuation index and its deviation from historical data exceed thresholds. This "spatiotemporal dual-check" mechanism ensures accurate identification of real and continuously escalating high-risk combinations, effectively avoiding misjudgments caused by short-term fluctuations or single-factor interference, and significantly improving the reliability of early warnings.

[0092] Figure 3 This is a schematic diagram of the high-risk combination judgment logic of a remote duty monitoring method according to an embodiment of this application, such as... Figure 3 As shown, the time series array is constructed using a circular buffer structure, with a preset sampling interval of 0.5 seconds and a buffer capacity of 1200 sampling points, corresponding to 10 minutes of monitoring data. Each sampling point contains data in five dimensions: smoke diffusion risk level code, door angle value, airflow intensity scalar, gap width measurement value, and average temperature. The risk level code is represented by integers, with high risk being 3, medium risk being 2, and low risk being 1.

[0093] Specifically, the implementation of the sliding time window involves two key parameters: window length and sliding step size. The window length is set to 60 sampling points, corresponding to a 30-second monitoring period; the sliding step size is 10 sampling points, meaning the window data is updated every 5 seconds. For the data sequence within the window, the number of times each parameter value changes is counted. Specifically, this is done by comparing the parameter values ​​of adjacent sampling points; when the difference exceeds a preset change threshold, the count is incremented by one. The threshold for door angle change is set to 2 degrees, airflow intensity change to 0.3 meters per second, gap width change to 0.5 millimeters, and temperature change to 1 degree Celsius. Dividing the number of changes by the number of sampling points within the window minus one yields the dynamic change frequency of each parameter. When the dynamic change frequency exceeds 0.3, it indicates that the parameter is in a state of rapid fluctuation.

[0094] It should be noted that the risk marker on the windward side is generated based on the real-time calculated angle between the door angle and the airflow velocity vector. When the angle is less than 30 degrees and the duration exceeds 5 seconds, a risk marker with a value of 1 is generated; otherwise, it is marked as 0. The determination of the small gap characteristic value is based on whether the gap width is within a small range of 3 to 8 millimeters; if it is within this range, the characteristic value is 1; otherwise, it is 0. These two binary markers serve as prerequisites for judging high-risk combinations.

[0095] For example, the specific application process of difference operation in temperature data processing is as follows: obtain the temperature value T at time i in the time series. i and the temperature value T at time i-1 i-1 Calculate the difference ΔT i =T i -T i-1 This yields a first-order difference sequence of temperatures. Performing a difference operation again on this first-order difference sequence produces a second-order difference sequence, reflecting the acceleration characteristics of temperature change. The rate of change of temperature distribution is defined as the ratio of the absolute value of the second-order difference to the sampling interval, expressed in degrees Celsius per second squared. When the rate of change of temperature distribution exceeds 0.5, it indicates the presence of a rapid temperature rise or fall.

[0096] This embodiment achieves continuous and stable recording of multi-dimensional environmental parameters through high-frequency sampling and a ring-shaped buffer structure. Based on a carefully set sliding window and changing threshold, the system can accurately quantify the dynamic fluctuation frequency of each parameter and identify persistent windward surfaces and small gaps. Furthermore, by calculating the acceleration of temperature changes, the system can keenly detect abnormal temperature rise trends. The synergistic effect of these specific parameters enables the system to reliably identify truly continuous and rapidly evolving "high-risk combinations" from the real-time data stream, significantly reducing misjudgments caused by instantaneous fluctuations or noise, thereby significantly improving the accuracy and timeliness of early warnings.

[0097] Preferably, the environmental parameter fluctuation index is calculated using a weighted summation method, with the weight coefficient of each parameter determined based on its influence on flue gas diffusion. The weight of the dynamic change frequency of the door angle is 0.3, the weight of the dynamic change frequency of the airflow intensity is 0.35, the weight of the dynamic change frequency of the gap width is 0.2, and the weight of the temperature distribution change rate is 0.15. The weighted summation value is divided by the sum of the weight coefficients to achieve normalization, resulting in a fluctuation index ranging from 0 to 1.

[0098] In one possible implementation, Pearson correlation analysis is used to identify the similarity between the current monitored data sequence and historical anomaly patterns. Normal operating data sequences from the same time period and season are extracted from the historical database as a baseline sequence, and the Pearson correlation coefficient between the current sequence and the baseline sequence is calculated. The correlation coefficient is calculated by dividing the covariance of the two sequences by the product of their respective standard deviations. When the correlation coefficient is less than 0.5, it indicates that the current sequence deviates from the normal pattern, and the degree of fluctuation anomaly is defined as 1 minus the absolute value of the correlation coefficient. An anomaly alarm is triggered when the degree of fluctuation anomaly exceeds 0.6.

[0099] For example, in a fire drill, the fire door was located at the end of a corridor, while the fire source was at the other end. As the fire spread, hot air flowed along the corridor, creating an airflow directed towards the fire door. Monitoring data showed that the angle between the door and the airflow was 15 degrees, generating a risk marker on the windward side; the gap width range was 5 millimeters, generating a small gap characteristic value; the temperature rapidly increased from 25 degrees Celsius to 45 degrees Celsius, with a temperature distribution change rate of 0.8; the dynamic change frequency of each parameter exceeded 0.4, and the calculated environmental parameter fluctuation index was 0.72; after comparison with historical normal data, the fluctuation anomaly was 0.75.

[0100] Understandably, the determination of high-risk combinations employs a multi-condition logical AND operation. Only when the risk marker on the windward side and the small gap characteristic value are both 1, and the environmental parameter fluctuation index exceeds the preset high-risk threshold of 0.6, while the fluctuation anomaly degree exceeds the anomaly threshold of 0.7, is the output of a true high-risk combination determination. This rigorous determination logic avoids false alarms caused by a single factor, improving the accuracy of fire risk identification. Furthermore, the high-risk combination determination result triggers the fire door's emergency response mechanism, including automatically closing the door, activating the water curtain sprinkler system, and sending alarm information to the fire control center. Through time series analysis and multi-dimensional parameter fusion, accurate monitoring and risk prediction of the complex environmental conditions surrounding the fire door are achieved.

[0101] S104. Based on the judgment results of high-risk combinations, extract feature data of similar scenarios from the pre-established historical case matching database, generate corresponding risk scenario labels, and determine the priority of handling.

[0102] Based on the high-risk combination assessment, feature vectors of currently monitored environmental parameters are extracted. These feature vectors include values ​​for five dimensions: windward surface marking, gap width, temperature distribution, airflow intensity, and smoke concentration. A cosine similarity algorithm is used to calculate the cosine of the angle between the current feature vector and the feature vectors of each case in the historical case database, yielding a similarity score. For historical cases with similarity scores exceeding a preset threshold, the corresponding fire development stage, smoke diffusion rate, and door damage level records are retrieved from the database. The fire type is coded based on the fire development stage, the concentration level is determined based on the smoke diffusion rate, and the temperature range is determined based on the door damage level. These three codes are combined to generate a risk scenario label. The risk scenario label is used to query a preset hazard level table and response timetable to obtain the corresponding hazard coefficient and time weight. A comprehensive score is calculated by multiplying the hazard coefficient by the time weight and adding a base score. The cases are then sorted from highest to lowest comprehensive score to determine the priority of response.

[0103] Specifically, in one implementation, the construction of the environmental parameter feature vector employs a normalization method, mapping the original values ​​of the five dimensions to a unified range of 0 to 1. The windward side is marked as a binary variable, directly taking either 0 or 1; the gap width value is normalized by dividing by the maximum value of 20 mm; the temperature distribution is divided by the standard value of 5 degrees Celsius per meter; the airflow intensity is divided by the upper limit value of 10 meters per second; and the smoke concentration is divided by the full-scale value of 1000 ppm.

[0104] Specifically, the calculation of cosine similarity involves vector dot product and modulus operations. Let the current feature vector be A, and the historical case feature vector be B. The similarity calculation formula is the dot product of the two vectors divided by the product of their moduli. When the similarity exceeds 0.85, the current scene is considered highly similar to a historical case, and detailed information about that historical case is extracted. The historical case database uses a hash index structure, with the hash value of the feature vector as the primary key, storing information such as the case occurrence time, duration, fire development stage, smoke diffusion rate, degree of damage to the door, response measures, and response effects.

[0105] It should be noted that the fire development stages are divided into four phases: initiation, development, intense fire, and decay, coded as 01, 02, 03, and 04 respectively. Smoke diffusion rate is categorized into three levels based on the distance diffused per second: less than 0.5 meters is low speed (coded L); 0.5 to 2 meters is medium speed (coded M); and greater than 2 meters is high speed (coded H). The degree of damage to doors is determined by the highest surface temperature: below 200 degrees Celsius is minor (coded 1); 200 to 400 degrees Celsius is moderate (coded 2); and above 400 degrees Celsius is severe (coded 3). These three codes are combined sequentially to form an eight-digit risk scenario label, such as 02M2 representing a scenario in the development phase, with medium-speed diffusion and moderate damage.

[0106] Preferably, the hazard factor is determined based on the fire development stage: 0.3 for the initial stage, 0.6 for the development stage, 1.0 for the intense stage, and 0.4 for the decay stage. The time weight reflects the urgency of the response, requiring a weight of 1.0 for responses within 5 minutes, 0.7 for responses within 10 minutes, and 0.4 for responses within 30 minutes. The overall score is calculated using linear weighting, with a base score of 50. The final score equals the base score plus the product of the hazard factor and the time weight, multiplied by 100.

[0107] In one possible implementation, when multiple similar historical cases are retrieved, the three cases with the highest similarity are selected, and their risk scenario tags are extracted. The final tags are determined through a voting mechanism. If the three tags are inconsistent, the tag of the case with the highest similarity takes precedence. The handling priority is divided into four levels based on the comprehensive score: scores above 140 are emergency, 100 to 140 are high-level, 60 to 100 are medium-level, and below 60 are low-level.

[0108] This embodiment uses a cosine similarity algorithm to intelligently match current high-risk scenarios with a historical case database, quickly identifying similar fire patterns. Based on the matching results, the system automatically generates a composite risk label that integrates fire stage, spread rate, and damage extent, and dynamically calculates the response priority according to the hazard level and response urgency. This process achieves a precise mapping from "current data" to "historical experience," enabling early warnings to no longer rely on a single threshold but rather on multi-dimensional assessments based on real cases. This significantly improves the accuracy of risk assessment and provides clear and quantifiable decision-making basis for the optimized allocation of emergency resources.

[0109] This embodiment achieves rapid and accurate matching of the current scenario with historical cases through normalization and cosine similarity calculation. It transforms complex multidimensional environmental parameters into standardized feature vectors and retrieves the most similar fire patterns from a historical database based on these vectors, thereby generating a composite risk label that integrates fire stage, spread rate, and damage extent. By combining hazard coefficients and response urgency to quantitatively calculate handling priorities, the system can intelligently distinguish between the severity and urgency of risks, thus transforming abstract data into clear action guidelines and significantly improving the decision-making efficiency of emergency command and the scientific nature of resource allocation.

[0110] S105. Based on the judgment result of high-risk combination, obtain real-time image data from the remote monitoring platform, perform boundary extraction technology on the door gap area in the image for analysis, identify the slight change trend of the gap width range, and determine whether there is a channel that accelerates the diffusion of smoke.

[0111] Based on the high-risk combination judgment results, real-time video streams of the fire door area are obtained from the remote monitoring platform. Continuous image frames are extracted according to a preset frame rate, and grayscale processing and histogram equalization are performed on the image frames to obtain preprocessed images. For the preprocessed images, the vertical and horizontal straight line features of the door frame are detected by Hough transform, and the area between adjacent vertical lines is determined as the region of interest (ROI) of the door gap. Within the ROI, the Canny edge detection algorithm is used to extract the door gap contour, obtaining a set of boundary pixels. Based on the boundary pixel set, the pixel distance between corresponding points on the left and right boundaries is calculated. Using a pre-calibrated conversion coefficient between pixels and actual distances, the pixel distance is converted into a physical width value. The width values ​​of each frame are stored in chronological order to construct a time-series width array. A fixed-length sliding window is used to traverse the time-series width array, and the rate of change of the width value within the window is calculated. If the rate of change of the width value shows an increasing trend in multiple consecutive windows and exceeds a preset diffusion threshold, the door gap area is marked as an accelerated smoke diffusion channel, indicating the existence of a high-risk diffusion path.

[0112] Specifically, in one implementation, the real-time video stream of the remote monitoring platform is acquired using a webcam, which is installed 3 to 5 meters away from the fire door, directly opposite it, with a field of view covering the entire door area. The video stream resolution is set to 1920×1080 pixels, and the frame rate is 25 frames per second. Based on the judgment result of high-risk combinations, the image acquisition module is triggered to extract key frames at a frequency of 5 frames per second to avoid redundant information between adjacent frames.

[0113] Specifically, grayscale processing employs a weighted average method, linearly combining the red, green, and blue channels of the color image with weight coefficients of 0.299, 0.587, and 0.114, respectively, to obtain a single-channel grayscale image. Histogram equalization remaps grayscale values ​​using a cumulative distribution function, evenly distributing pixel values ​​concentrated in a certain grayscale range in the original image across the entire grayscale range, thus enhancing the contrast between the door gap area and the background. After preprocessing, the boundary features of the door gap exhibit a clear transition between light and dark areas in the grayscale image.

[0114] It should be noted that the implementation of Hough transform for line detection involves parameter space transformation. A straight line in image space is represented in polar coordinate parameter space as ρ = x × cosθ + y × sinθ, where ρ is the perpendicular distance from the origin to the line, and θ is the angle between the perpendicular and the x-axis. By traversing θ from 0 to 180 degrees, the ρ value corresponding to each edge point is calculated and accumulated in the parameter space. When the accumulated value exceeds a preset threshold, it is identified as a straight line. Fire door frames typically present four main straight lines: two vertical lines corresponding to the left and right door frames, and two horizontal lines corresponding to the top and bottom door frames. The area between the vertical lines is the location of the door seam. By calculating the difference in x-coordinate between the vertical lines and the difference in y-coordinate between the horizontal lines, the boundary coordinates of the rectangular region of interest are determined.

[0115] For example, the application of the Canny edge detection algorithm in door seam contour extraction involves multiple processing stages. A Gaussian filter (5×5 size, standard deviation 1.4) is used to smooth the region of interest. The amplitude and direction of the image distribution are calculated, and the Sobel operator is used to calculate the distribution components in the horizontal and vertical directions respectively. Non-maximum suppression compares pixel values ​​along the distribution direction, retaining local maxima and suppressing other points. A dual-thresholding process sets a high threshold at the 0.7 quantile of the distribution amplitude and a low threshold at 0.4 times the high threshold; pixels above the high threshold are identified as strong edges, and those between the two thresholds are identified as weak edges. Through connectivity analysis, weak edges connected to strong edges are retained, while isolated weak edges are removed, ultimately resulting in a continuous door seam contour.

[0116] Preferably, the conversion from pixel distance to physical distance relies on the prior acquisition of camera calibration parameters. The calibration process uses a checkerboard calibration board. A calibration board of known dimensions is placed on the plane of the fire door, and calibration images are captured from multiple angles using the camera. The Zhang Zhengyou calibration method is used to calculate the camera's intrinsic parameter matrix and distortion coefficients. The intrinsic parameter matrix includes parameters such as focal length and principal point coordinates. A world coordinate system is established on the door seam plane, and the image coordinates are mapped to world coordinates using a perspective transformation matrix. After calibration, the conversion coefficient between pixels and actual distance is obtained, typically corresponding to a physical distance of 0.2 to 0.5 millimeters per pixel.

[0117] In one possible implementation, corresponding points on the left and right boundaries of the boundary pixel set are determined by y-coordinate matching. For each y-coordinate value, the corresponding x-coordinate is found on both the left and right boundaries, and the Euclidean distance between the two points is calculated as the pixel value of the door gap width at that height. The pixel width is multiplied by a conversion factor to obtain the actual physical width value in millimeters.

[0118] Understandably, the time series width array is constructed using a circular buffer storage structure with a capacity of 300 data points, corresponding to 60 seconds of historical data. Each time a new width value is obtained, it is added to the end of the buffer, while the oldest data point is removed. The sliding window length is set to 30 data points, corresponding to a 6-second time period, with a sliding step of 5 data points, meaning the window position is updated once per second. Furthermore, the width change rate is calculated using the least squares method to fit the data points within the window. Let the time series be t, and the width value series be w. A linear regression yields the fitted line w = at + b, where the slope 'a' is the width change rate in millimeters per second. When the change rate of five consecutive windows is positive and shows an increasing trend, it is determined that the width is continuously increasing. The preset diffusion threshold is set according to fire door design standards, typically 0.5 millimeters per second.

[0119] For example, in a fire test, the initial door gap width was 5 mm. As the door deformed due to heat, the width gradually increased. Image analysis detected that the width was 5.5 mm at the 10th second, 6.2 mm at the 15th second, and 7.1 mm at the 20th second, with calculated rates of change of 0.1, 0.14, and 0.18 mm per second, respectively, showing an accelerating trend. When the rate of change exceeded the threshold of 0.5 mm per second, the system determined that the door gap had formed an accelerated smoke diffusion channel and immediately triggered protective measures, including activating the door gap sealing device and pressurized air supply system to prevent smoke from spreading to a safe area.

[0120] This embodiment upgrades video surveillance from "visible" to "measurable" through image processing technology. It utilizes Hough transform and Canny edge detection to accurately locate and extract the contour of the door gap. Through pre-calibration, pixel widths are converted into physical dimensions, enabling continuous and accurate measurement of the door gap width. Furthermore, by analyzing the rate of change of width over time, the system can keenly identify the critical risk signal that the door gap is "accelerating" its expansion. This provides intuitive and quantitative image evidence for determining an "accelerated smoke diffusion channel," which, when corroborated by sensor data, constitutes a multi-dimensional, highly reliable risk confirmation mechanism, greatly reducing reliance on a single data source and the possibility of misjudgment.

[0121] S106. Based on the judgment results of the accelerated flue gas diffusion channel, the airflow interaction intensity and smoke concentration readings are combined to determine the high-risk scenario judgment results.

[0122] Based on the assessment of the accelerated smoke diffusion channel, the current airflow interaction intensity and smoke concentration readings are obtained from the sensor network, and these two values ​​are combined to form a real-time monitoring data pair. This real-time monitoring data pair is then matched against a pre-established two-dimensional risk assessment table. The table's row index represents the airflow intensity range, the column index represents the smoke concentration range, and the elements within the table are risk level codes. When both the airflow interaction intensity and the smoke concentration exceed a preset intensity threshold, the scenario is determined to be high-risk.

[0123] This embodiment establishes a two-dimensional risk assessment table for airflow intensity and smoke concentration, rapidly combining and matching direct readings of these two key parameters to achieve immediate diagnosis of high-risk scenarios. The core advantage of this method lies in its high efficiency and clarity of decision-making: the system does not require complex calculations; it only needs to compare real-time monitoring data with preset thresholds and matrices to immediately trigger a high-risk assessment when both airflow and smoke concentration exceed safety limits. This dual-threshold linkage mechanism effectively avoids misjudgments caused by a single parameter anomaly, significantly improving the accuracy and reliability of the final risk assessment result, and providing a clear and decisive basis for subsequently initiating the highest-level emergency response.

[0124] Specifically, in one implementation, the airflow interaction intensity is measured by a wind speed sensor in meters per second, and the smoke concentration is acquired by a photoelectric smoke sensor in milligrams per cubic meter. When an accelerated smoke diffusion channel is detected, a data fusion determination process is triggered.

[0125] Specifically, the two-dimensional risk assessment table adopts a 5×5 matrix structure. The row index corresponds to five airflow intensity ranges: 0-2, 2-4, 4-6, 6-8, and greater than 8 meters per second; the column index corresponds to five smoke concentration ranges: 0-50, 50-150, 150-300, 300-500, and greater than 500 milligrams per cubic meter. Elements within the table use a three-level risk coding system: 1 for low risk, 2 for medium risk, and 3 for high risk.

[0126] It should be noted that the preset intensity threshold is set to 4 meters per second, and the concentration threshold is set to 150 milligrams per cubic meter. When both the airflow interaction intensity and the smoke concentration exceed their respective thresholds, the risk level obtained from the table is usually 3, which is directly judged as a high-risk scenario. If only one of them exceeds the threshold, the risk level is determined by looking up the table based on the specific numerical range.

[0127] Preferably, the high-risk scenario determination result triggers the fire door's emergency response mechanism, including automatic closing, activation of water sprinkler and audible and visual alarms, to achieve effective early fire control.

[0128] S107. If a high-risk scenario is identified, the system retrieves the corresponding handling protocol sequence from the database, which corresponds to the handling priority and risk scenario label, generates a real-time response instruction, and automatically adjusts the fire door status to obtain an optimized safety configuration status.

[0129] If a high-risk scenario is identified, a query index is constructed based on priority and risk scenario tags. Matching records are retrieved from a pre-established handling protocol database to obtain a protocol sequence containing door angle adjustment values, sealing strip inflation pressure values, and smoke exhaust valve opening percentages. The parameter values ​​in the protocol sequence are converted: the door angle adjustment value is multiplied by the number of pulses per unit angle of the servo motor to obtain the total number of control pulses; the sealing strip inflation pressure value is divided by the air pump's air supply per unit time to obtain the working duration; and the smoke exhaust valve opening percentage is multiplied by the stepper motor's full stroke steps to obtain the target number of steps. These are combined to form a real-time response command set. This real-time response command set is sent to the door servo motor, sealing strip air pump, and smoke exhaust valve stepper motor via a pre-defined communication interface. The position, pressure, and opening signals returned by each actuator are monitored. When the position signal reaches the target angle, the pressure signal reaches the set pressure, and the opening signal reaches the target opening, the optimized safety configuration status is confirmed.

[0130] This embodiment quickly matches preset handling protocols based on priority and scene tags, and accurately converts the safety parameters into drive commands for each execution component. By issuing commands through a standardized communication interface and monitoring feedback signals in real time, the system achieves coordinated control of door posture adjustment, sealing enhancement, and smoke exhaust linkage, ultimately enabling the fire door to reach the predetermined optimized safety state. This process forms a complete automated response chain of "intelligent decision-making - precise execution - closed-loop verification," ensuring that matching protective actions can be executed immediately and reliably after high-risk confirmation, thereby greatly improving the timeliness and accuracy of fire emergency response.

[0131] Specifically, in one implementation, the disposal protocol database adopts a relational database structure, containing four main fields: protocol number, applicable scenario, priority level, and execution parameters. The query index is composed of a combination of priority values ​​and hash values ​​of risk scenario tags, enabling fast and accurate matching. Each protocol record contains three sets of execution parameters: the door angle adjustment value ranges from 0 to 90 degrees, the sealing strip inflation pressure value ranges from 0.1 to 0.5 MPa, and the smoke exhaust valve opening percentage ranges from 0 to 100%.

[0132] Specifically, the parameter conversion process is calculated based on the physical characteristics of each actuator. The number of pulses per unit angle for the servo motor is determined by the motor model, typically 100 pulses per degree; therefore, a 30-degree adjustment angle corresponds to 3000 control pulses. The air supply of the sealing strip air pump is 0.01 MPa per second; if the target pressure is 0.3 MPa, the working time is calculated to be 30 seconds. A stepper motor controlling the smoke exhaust valve from fully closed to fully open requires 2000 steps; a 50% opening corresponds to a target step count of 1000 steps.

[0133] It should be noted that the real-time response instruction set uses a hexadecimal encoding format. Each instruction consists of four parts: a device address code, a function code, a data field, and a checksum. The device address code identifies different execution units, the function code defines the operation type, the data field carries specific control parameters, and the checksum is calculated using a cyclic redundancy check algorithm. The communication interface uses a serial communication protocol with a baud rate of 9600, 8 data bits, and 1 stop bit.

[0134] Preferably, the monitoring of the feedback signals adopts a polling mechanism, querying the status of each actuator every 100 milliseconds. The position signal obtains the current angle value of the door through an encoder, compares it with the target angle, and a deviation within ±2 degrees is considered to be in position. The pressure signal measures the internal pressure of the sealing strip in real time through a pressure sensor; inflation is confirmed to be complete when it reaches 95% or more of the target pressure. The opening signal is determined by the stepper motor's step counter; valve adjustment is confirmed to be complete when the actual number of steps matches the target number.

[0135] In one possible implementation, the optimized safety configuration includes three aspects: the door is at a specified closing angle, the sealing strip is inflated to form an effective seal, and the smoke exhaust system is activated as needed. Through the synergistic effect of these multiple protective measures, the spread of fire smoke can be effectively controlled, ensuring the safety of personnel evacuation routes.

[0136] S108. Based on the optimized safety configuration status, update the dynamic change frequency record in the monitoring log, and at the same time transmit the adjusted environmental interaction parameter data to the central processing unit, cyclically verify the attenuation trend of airflow interaction intensity, and determine whether the overall risk has been reduced to a controllable range.

[0137] Based on the optimized safety configuration, the current door angle and sealing pressure are written into the monitoring log. The time change rate between the values ​​and the previous recorded values ​​is calculated to obtain the dynamic change frequency record of each parameter. The adjusted door angle, airflow intensity, gap width, temperature, and smoke concentration are re-collected by sensors to form adjusted environmental interaction parameters. The dynamic change frequency record and environmental interaction parameters are packaged into a data frame and transmitted to the central processing unit via the network interface. After receiving the confirmation signal from the central processing unit, the airflow interaction intensity value is read cyclically at preset time intervals. The difference between the current intensity value and the previous value is calculated. If the intensity value read multiple times shows a decreasing trend and the decreasing rate gradually increases, it is determined that the airflow interaction intensity is attenuating. When the intensity value drops below the preset safety threshold, it is determined that the overall risk has been reduced to a controllable range.

[0138] This embodiment employs an "execution-verification" closed-loop design, immediately re-collecting environmental parameters after adjusting the fire door status and continuously tracking the attenuation trend of airflow intensity. By analyzing whether the intensity value continuously decreases and eventually falls below the safety threshold, the system can dynamically and quantitatively assess the actual effectiveness of protective measures, thereby scientifically determining whether the overall risk has been controlled. This not only ensures the integrity of the emergency response but also guarantees that the system will not prematurely terminate the response before the danger has been eliminated, significantly improving the reliability and precision of safety management.

[0139] Figure 4 This is a schematic diagram illustrating the risk attenuation trend of a remote duty monitoring method according to an embodiment of this application. Figure 4 As shown, the monitoring log uses a timestamp-indexed database structure. Each record contains four fields: recording time, door angle, sealing pressure, and smoke exhaust opening. The dynamic change frequency is calculated based on the time interval and parameter difference between two adjacent records. The change rate is obtained by dividing the difference by the time interval, with the unit being the change per second. When the change rate exceeds a preset threshold, it is marked as a rapid change state.

[0140] Specifically, the adjusted environmental interaction parameters are acquired by reactivating the sensor network. Door angle sensors, anemometers, laser rangefinders, infrared thermometers, and smoke concentration sensors simultaneously collect current data, forming a five-dimensional parameter vector. Data frames are packaged in a fixed format, including a frame header, timestamp, dynamic frequency record, environmental interaction parameters, and checksum, with a total length of 128 bytes. The network interface uses the Ethernet protocol with a transmission rate of 100 megabits per second.

[0141] It should be noted that the cyclic verification process is set to execute once every 5 seconds, with a maximum of 20 loops. Each loop reads the current airflow interaction intensity value and stores it in the loop buffer. The attenuation trend is determined based on data from 5 consecutive sampling points. By calculating the difference sequence between adjacent points, if all differences are negative and their absolute values ​​show an increasing trend, attenuation is confirmed. The magnitude of the decrease is obtained by calculating the second difference of the difference sequence; a negative second difference indicates an increasing magnitude of the decrease.

[0142] Preferably, the preset safety threshold is determined according to the design specifications of the fire door, and is typically set to 2 meters per second. When the airflow interaction intensity drops below this threshold and remains below the threshold for three consecutive sampling cycles, the risk is determined to be within a controllable range. After receiving the risk controllable determination result, the central processing unit records the current system status and stops the cyclic verification process.

[0143] In one possible implementation, if a rebound in airflow intensity is detected during the cyclic verification process, the current cycle is immediately interrupted, and the fire door adjustment procedure is re-executed. By monitoring real-time changes in airflow intensity, protective measures are dynamically adjusted to achieve continuous control over fire risks.

[0144] For example, after a fire emergency response, the initial airflow intensity was 8 meters per second. After adjustment with fire doors, the first cycle measured 7.2 meters per second, the second 6.3 meters per second, and the third 5.2 meters per second, with decreasing rates of 0.8, 0.9, and 1.1 respectively, showing an increasing trend. This confirmed that the system's protective measures were effective, and the cycle continued until the intensity dropped below 2 meters per second.

[0145] The above embodiments, by simultaneously collecting and fusing multi-dimensional data such as door orientation, airflow, gaps, temperature, and smoke concentration, combined with time series analysis, have for the first time achieved accurate identification of the dynamic high-risk combination of "windward side" and "minor gaps." Furthermore, by verifying the diffusion channel through image analysis and generating intelligent response strategies based on historical case matching, they have achieved integrated decision-making from risk perception and dual verification to precise response. Finally, by continuously tracking risk attenuation through a closed-loop verification mechanism, fire doors are upgraded from passive isolation facilities to active safety systems with dynamic early warning, intelligent response, and adaptive control capabilities. This fundamentally solves the problems of missed detection, misjudgment, and delayed response of traditional methods for specific high-risk situations, significantly improving the accuracy, timeliness, and reliability of fire smoke diffusion control.

[0146] This invention provides a remote duty monitoring system, such as... Figure 2 As shown, the system mainly includes:

[0147] The data acquisition and preprocessing module is used to collect data on the door's orientation angle, airflow velocity vector, airflow interaction intensity, gap width range, temperature distribution, and smoke concentration in real time through a multi-point sensor array deployed around the fire door. The module then performs noise reduction processing on the collected raw data to obtain environmental interaction parameters.

[0148] The risk level analysis module is used to perform geometric analysis on the angle between the door orientation angle and the airflow velocity vector based on environmental interaction parameters, and combined with the gap width range to determine the potential smoke diffusion risk level.

[0149] The high-risk combination judgment module is used to track and analyze the fluctuations of environmental interaction parameters by combining the potential flue gas diffusion risk level with the dynamic change frequency and temperature distribution, and using the time series comparison method to determine whether there is a high-risk combination of windward risk marker and small gap characteristic value superposition.

[0150] The priority determination module is used to extract feature data of similar scenarios from a pre-established historical case matching database based on the judgment results of high-risk combinations, generate corresponding risk scenario labels, and determine the priority of disposal.

[0151] The image analysis module is used to acquire real-time image data from the remote monitoring platform based on the judgment results of high-risk combinations. It performs boundary extraction technology on the door gap area in the image for analysis, identifies the subtle change trend of the gap width range, and determines whether there is a channel that accelerates the diffusion of smoke.

[0152] The high-risk scenario determination module is used to determine the high-risk scenario determination result by combining the judgment result of the accelerated flue gas diffusion channel with the airflow interaction intensity and smoke concentration readings.

[0153] The response execution module is used to retrieve the handling protocol sequence corresponding to the handling priority and risk scenario label from the database if the scenario is determined to be high-risk, generate real-time response instructions, automatically adjust the status of the fire door, and obtain an optimized safety configuration status.

[0154] The status update and verification module is used to update the dynamic change frequency record in the monitoring log according to the optimized safety configuration status, and at the same time transmit the adjusted environmental interaction parameter data to the central processing unit to cyclically verify the attenuation trend of airflow interaction intensity and determine whether the overall risk has been reduced to a controllable range.

[0155] This embodiment ensures the integrity and timeliness of status verification data through standardized data encapsulation and high-speed network transmission. Its core value lies in introducing a continuous and intelligent "effect evaluation" cycle: the system not only checks whether the airflow intensity has dropped to the threshold, but also analyzes its continuously decreasing acceleration trend to predict in advance whether the protective measures remain effective. Once a trend reversal is detected, the system can immediately interrupt the verification and restart the adjustment process, thereby achieving a continuous risk management closed loop of dynamic tracking, intelligent judgment, and autonomous iteration. This significantly improves the system's adaptive control capabilities and long-term reliability in complex fire environments.

[0156] This embodiment of the system, through modular design, seamlessly integrates data acquisition, risk analysis, image verification, intelligent decision-making, and closed-loop control, forming a complete automated management and control chain of "perception-assessment-response-verification." It integrates multi-source sensor data, real-time image analysis, and historical case matching to achieve integrated intelligent response from risk identification to precise handling. Its core value lies in upgrading traditional, decentralized, and passive fire door monitoring into a dynamic, proactive, and self-verifiable intelligent safety system through system-level collaboration, significantly improving the overall effectiveness, timeliness, and reliability of building fire smoke diffusion control.

[0157] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A remote duty monitoring method, characterized in that, The method includes: By deploying a multi-point sensor array around the fire door, data such as the door's orientation angle, airflow velocity vector, airflow interaction intensity, gap width range, temperature distribution, and smoke concentration are collected in real time. The raw data is then denoised to obtain environmental interaction parameters. Geometric analysis is performed on the angle between the door orientation angle and the airflow velocity vector based on environmental interaction parameters, and the potential smoke diffusion risk level is determined by combining the gap width range. By combining the potential flue gas diffusion risk level with the dynamic change frequency and temperature distribution, the fluctuation of environmental interaction parameters is tracked and analyzed using the time series comparison method to determine whether there is a high-risk combination of windward risk marker and small gap characteristic value superposition. Based on the high-risk combination judgment results, similar scenario feature data are extracted from the historical case matching database to generate risk scenario labels and determine priorities; Based on the high-risk combination judgment results, real-time image data from the remote monitoring platform is obtained. Boundary extraction analysis is performed on the door gap area to identify the subtle change trend of the gap width and determine whether there is an accelerated flue gas diffusion channel. The high-risk scenario assessment result is determined by combining the results of the accelerated flue gas diffusion channel assessment with the airflow interaction intensity and smoke concentration readings. When a high-risk scenario is identified, a real-time response instruction is generated by retrieving the handling protocol sequence corresponding to the priority and risk scenario label from the database, and the fire door status is automatically adjusted to obtain an optimized safety configuration status. Based on the optimized safety configuration status, the dynamic change frequency record in the monitoring log is updated and the adjusted environmental interaction parameters are transmitted to the central processing unit. The airflow interaction intensity attenuation trend is verified cyclically to determine whether the overall risk has been reduced to a controllable range. The process involves combining potential flue gas diffusion risk levels with dynamic change frequency and temperature distribution, and using time series comparison methods to track and analyze fluctuations in environmental interaction parameters to determine whether there is a high-risk combination of windward risk markers and small gap characteristic values, including: Environmental interaction parameters are continuously collected based on the potential flue gas diffusion risk level, and a time series array is constructed. The dynamic change frequency is obtained by using the sliding time window method to count the ratio of the number of changes of each parameter to the length of the time window. When the angle between the door angle and the airflow velocity vector is less than the preset value, a risk mark is generated on the windward side. When the gap width is in a small range, a small gap feature value is generated. The change in temperature distribution is obtained by differential operation, the rate of change in temperature distribution is calculated, and the environmental parameter fluctuation index is obtained by weighted summation of dynamic change frequency and temperature distribution change rate and normalization. The fluctuation anomaly degree is obtained by comparing with historical data series of the same period. When the windward risk marker and the small gap characteristic value coexist, and the environmental parameter fluctuation index and fluctuation anomaly degree both exceed the corresponding thresholds, a high-risk combination is determined to exist.

2. The remote duty monitoring method according to claim 1, characterized in that, The system uses a multi-point sensor array deployed around the fire door to collect data in real time on the door's orientation angle, airflow velocity vector, airflow interaction intensity, gap width range, temperature distribution, and smoke concentration readings. The raw data is then denoised to obtain environmental interaction parameters, including: A sensor array is arranged in a regular hexagonal grid around the fire door. The sensor network topology and spatial coordinates are obtained by triangulation based on the known distance and angle sensor readings between adjacent nodes. Simultaneously collect tilt angle data, airflow velocity vector, airflow interaction intensity, temperature distribution data, gap width range data, and smoke concentration readings. Perform coordinate transformation on the airflow velocity vector to calculate the cosine value of the angle between the door normal vector and the airflow velocity vector and determine whether it is a windward state. If the condition is determined to be windward, a Kalman filter is used to suppress noise in various data, and the filtered data is fused by a weighted average method to obtain the environmental interaction parameters.

3. The remote duty monitoring method according to claim 1, characterized in that, The process involves geometrically analyzing the angle between the door's orientation and the airflow velocity vector based on environmental interaction parameters, and combining this with the gap width range to determine the potential smoke diffusion risk level, including: Extract the door orientation angle and airflow velocity vector from the environmental interaction parameters, obtain the cosine value of the included angle through vector dot product operation, and calculate the actual included angle value using the inverse cosine function; The corresponding permeability coefficient value is retrieved from a pre-established flue gas permeability coefficient database, and the potential flue gas diffusion risk level is determined by comparing the flue gas permeability coefficient with a preset risk threshold set.

4. The remote duty monitoring method according to claim 1, characterized in that, The step of extracting similar scenario feature data from the historical case matching database based on the high-risk combination judgment results to generate risk scenario labels and determine priorities includes: Extract the feature vector of the current environment parameters, and calculate the similarity value with the feature vector of historical cases using the cosine similarity algorithm; Read historical case records of fire development stages, smoke diffusion rates, and door damage levels that exceed the similarity threshold, and generate risk scenario labels; By querying the preset hazard level table and response timetable using risk scenario tags, a comprehensive score is calculated to determine the priority of handling.

5. The remote duty monitoring method according to claim 1, characterized in that, The process of obtaining real-time image data from the remote monitoring platform based on the high-risk combination judgment result, performing boundary extraction analysis on the door gap area to identify subtle changes in gap width, and determining whether there is an accelerated flue gas diffusion channel includes: Continuous image frames are extracted from the real-time video stream of the fire door area obtained from the remote monitoring platform, and then grayscale and histogram equalization are performed to obtain the preprocessed image; The region of interest for the door gap is determined by detecting the straight line features of the door frame using Hough transform, and the door gap contour is extracted by Canny edge detection within the region of interest to obtain the set of boundary pixels. Calculate the pixel distance between the left and right boundaries and convert it into a physical width value. Construct a time series width array in chronological order. Use a sliding window to traverse and calculate the rate of change of the width value within the window. When the rate of change of the width shows an increasing trend in multiple consecutive windows, it is marked as an accelerated flue gas diffusion channel.

6. The remote duty monitoring method according to claim 1, characterized in that, The process of determining the high-risk scenario based on the accelerated smoke diffusion channel judgment result, combined with the airflow interaction intensity and smoke concentration readings, includes: The data obtained from the sensor network, including airflow interaction intensity and smoke concentration readings, are combined to form real-time monitoring data pairs, which are then matched with a pre-established two-dimensional risk assessment table. When the airflow interaction intensity exceeds a preset intensity threshold and the smoke concentration exceeds a preset concentration threshold, it is determined as a high-risk scenario.

7. The remote duty monitoring method according to claim 1, characterized in that, When a scenario is determined to be high-risk, a real-time response instruction is generated by retrieving the handling protocol sequence corresponding to the priority and risk scenario label from the database, and automatically adjusting the fire door status to obtain an optimized safety configuration status, including: If a scenario is determined to be high-risk, a query index is built based on priority and risk scenario tags to retrieve the protocol sequence from the handling protocol database. The parameter values ​​in the protocol sequence are converted to form a real-time response instruction set. The real-time response instruction set is sent to the execution unit through a preset communication interface, the return signal is monitored, and the optimized security configuration status is confirmed.

8. The remote duty monitoring method according to claim 1, characterized in that, The step of updating the dynamic change frequency record in the monitoring log according to the optimized security configuration status and transmitting the adjusted environmental interaction parameters to the central processing unit, and cyclically verifying the attenuation trend of airflow interaction intensity, includes: Based on the optimized safety configuration status, the current door angle and sealing pressure are written into the monitoring log, and the time change rate is calculated to obtain the dynamic change frequency record. The dynamically changing frequency records and the re-acquired and adjusted environmental interaction parameters are packaged and transmitted to the central processing unit. The airflow interaction intensity values ​​are read cyclically to calculate the difference. When the intensity value continuously decreases and drops below the safety threshold, it is determined that the overall risk has been reduced to a controllable range.

9. A remote duty monitoring system, implemented based on the remote duty monitoring method according to claim 1, characterized in that, The system includes: The data acquisition and preprocessing module is used to collect preliminary environmental interaction parameters in real time through a multi-point sensor array around the fire door, and generate environmental interaction parameters after noise reduction. The risk level analysis module is used to analyze the angle between the door and the airflow and the gap width based on environmental interaction parameters to determine the risk level of flue gas diffusion. The high-risk combination judgment module is used to combine risk level, dynamic change frequency and temperature data, track environmental parameter fluctuations through time series comparison, and determine whether there is a high-risk combination of windward surface and small gap; The priority determination module is used to extract similar scenario features from the historical case database based on high-risk combination results, generate risk scenario labels, and determine the priority of disposal. The image analysis module is used to acquire real-time images from remote monitoring based on high-risk combination results, perform boundary extraction analysis on the door gap area, identify subtle changes in gap width, and determine whether an accelerated flue gas diffusion channel exists. The high-risk scenario determination module is used to determine high-risk scenarios by combining the results of the accelerated smoke diffusion channel judgment with the airflow intensity and smoke concentration. The response execution module is used in high-risk scenarios to retrieve the handling protocols corresponding to the priority and risk scenario tags, generate real-time instructions, and automatically adjust the fire door to a safe configuration state. The status update and verification module is used to update the dynamic change frequency of the monitoring log according to the security configuration status, transmit the adjusted environmental interaction parameters, cyclically verify the airflow intensity attenuation trend, and determine whether the overall risk is controllable.

Images