A fire point location system integrating aspiration detection and pipeline airflow analysis

Abstract

This invention discloses a fire point location system that integrates aspirating detection and pipeline airflow analysis, comprising: an aspirating sensing module, which collects smoke concentration and dynamic airflow parameters in real time through an air sampling pipeline network; a pipeline airflow analysis module, which constructs a pipeline airflow analysis model based on fluid dynamics equations, performs calculations, and outputs the suspected fire area; an airflow field compensation and correction module, which performs reverse compensation and correction on the suspected fire area based on a preset dynamic airflow path diagram and / or the real-time calculated smoke drift distance, and outputs the corrected fire area; and a precise fire point location module, which calls the detection data from the location sub-module within the corrected area, integrates and judges, and outputs the precise coordinates of the fire point. This invention eliminates the disturbance effect of a fixed airflow field on the smoke diffusion path by integrating pipeline airflow analysis and airflow field drift compensation, achieving precise location from the regional level to the equipment level, and supporting both dynamic diagram rapid correction and real-time data flexible correction modes.

Description

Technical Field

[0001] This invention relates to the field of electrical fire monitoring technology, and more specifically to a fire ignition point location system that integrates aspiration detection and pipeline airflow analysis. Background Technology

[0002] Due to its high sensitivity and early warning capabilities, aspirating smoke detection technology is widely used in the protection of precision equipment such as computer rooms, energy storage stations, power distribution cabinets, and offshore drilling platforms. This type of system continuously extracts air samples through an air sampling pipeline network deployed in the protected area and delivers them to a central detector for analysis, enabling it to issue early warnings when the fire is in the smoldering stage and before visible flames have appeared.

[0003] In the prior art, solutions integrating aspirating fire detection with automatic fire extinguishing functions have emerged. For example, the invention patent application with publication number CN115645784 A discloses an aspirating detection automatic fire extinguishing device, which realizes the linkage control of detection and fire extinguishing. However, this solution relies on a single detection parameter for fire judgment, resulting in a high false alarm rate. Furthermore, it does not have a dedicated fire location module and can only achieve full-area coverage fire extinguishing, which poses a risk of wasting extinguishing agents and causing secondary damage to precision equipment in non-fired areas. The utility model patent with publication number CN219758911U uses a GPS location module to achieve fire location, but the GPS positioning accuracy is low and it is difficult to meet the precise positioning needs of indoor, small protected areas or cabinet-level equipment. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a fire point location system that integrates aspirating detection with pipeline airflow analysis. The goal is to solve the problem of low fire point location accuracy in existing aspirating fire detection systems. Specifically, existing technologies rely on the position of smoke reaching each sampling hole for estimation, but fail to consider the disturbance effect of fixed airflow fields generated by environmental control equipment such as air conditioners and fresh air systems within the protected area on the natural diffusion path of smoke. This results in a location deviation between the estimated suspected fire area and the actual fire point.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A fire ignition point location system integrating aspiration detection and pipeline airflow analysis includes: The aspirating sensing module collects multi-dimensional sensing data in real time, including at least smoke concentration parameters and airflow dynamic parameters, through aspirating detection devices deployed in the air sampling pipeline network. The pipeline airflow analysis module calculates the multi-dimensional sensing data through a preset pipeline airflow analysis model, generating airflow distribution ratios, smoke particle transmission delays, and pipeline resistance characteristic parameters for each sampling branch. Based on the calculation results, it inverts the smoke source area and outputs the suspected fire area. The airflow field compensation and correction module performs reverse compensation and correction of smoke drift in the suspected fire area based on the preset dynamic airflow direction diagram and / or the smoke drift distance calculated from the multi-dimensional sensing data of each sampling point, and outputs the corrected fire area. The precise fire point location module calls upon the detection data from the location sub-modules deployed in various devices within the fire area, fuses and judges the smoke concentration parameters fed back by the location sub-modules, and outputs the precise coordinate information of the fire point.

[0007] Furthermore, the dynamic airflow direction diagram includes the airflow direction vector and airflow velocity scalar at each location in space. The calibration distance is calculated based on the airflow velocity scalar and the estimated residence time of smoke particles in the airflow field. The suspected fire area is shifted along the opposite direction of the airflow by the calibration distance to obtain the corrected fire area.

[0008] Furthermore, the airflow compensation and correction module includes an estimated residence time calculation strategy. The estimated residence time calculation strategy includes determining the spatial location of each sampling point within the suspected fire area, associating and matching each sampling point with the corresponding airflow direction vector and airflow velocity scalar in the airflow direction dynamic diagram, obtaining the actual arrival time of smoke particles from the fire source to each sampling point, calculating the theoretical transmission time based on the spatial distance between each sampling point and the center of the suspected fire area and the corresponding airflow velocity scalar, and using the difference between the actual arrival time and the theoretical transmission time as the estimated residence time of smoke particles in the airflow field.

[0009] Furthermore, the airflow compensation and correction module includes a smoke drift distance calculation strategy. This strategy includes obtaining the smoke concentration parameters of each sampling point within the suspected fire area within the same time window, calculating the smoke concentration difference between adjacent sampling points, and constructing a spatial gradient distribution field of smoke concentration. Based on the direction of the fastest decrease in concentration value in the smoke concentration gradient distribution field, and combined with the arrival time sequence of smoke particles at each sampling point, the actual drift direction of the smoke particles is determined. Then, based on the actual drift direction and the spatial distance between each sampling point, the distance from the fire source to the sampling point where the smoke was first captured is calculated as the smoke drift distance.

[0010] Furthermore, the pipeline airflow analysis module includes a pipeline airflow analysis model construction strategy. The pipeline airflow analysis model construction strategy includes a mathematical framework for constructing a pipeline airflow analysis model based on the fluid mechanics continuity equation, Bernoulli's equation, and pipeline resistance characteristics. The pipeline resistance characteristics include the friction resistance calculated based on the friction coefficient, pipe length, pipe diameter, and average airflow velocity in the pipe, as well as the local resistance calculated based on the local resistance coefficient and the average airflow velocity in the pipe. Based on the preset pipeline topology, the number of branches, the length of each branch, the diameter of each branch, the local resistance coefficient of each branch, the number of each sampling hole, and the spatial location of each sampling hole are input into the mathematical framework to generate a pipeline airflow analysis model that includes a total flow distribution model and a smoke particle transmission delay model. The total flow distribution model is used to calculate the airflow distribution ratio of each branch and the inlet flow rate and velocity of each sampling hole based on the total flow resistance of each branch. The smoke particle transmission delay model is used to calculate the transmission delay required for smoke particles to travel from the sampling hole to the detection host.

[0011] Furthermore, the pipeline airflow analysis module also includes a real-time pipeline status calculation strategy. This strategy includes calculating the airflow distribution ratio of each branch, the inlet flow rate and velocity of each sampling hole, the pressure loss along the pipeline, and the local pressure loss based on the total airflow value, total negative pressure value, and pipeline topology of the real-time acquisition and detection host, using the total flow distribution model and the flow resistance balance equation of each branch. Based on the pipe length and the average airflow velocity in each branch, the smoke particle transmission delay of each sampling branch is calculated using the smoke particle transmission delay model.

[0012] Furthermore, the pipeline airflow analysis module also includes a fire-suspect area determination strategy. The fire-suspect area determination strategy includes eliminating invalid sampling branches based on the airflow and velocity of each sampling hole, determining candidate smoke source branches based on the smoke particle transmission delay among the remaining valid sampling branches, verifying high confidence based on the airflow distribution ratio of the candidate smoke source branches, and determining the spatial coverage area corresponding to the verified candidate smoke source branches as the fire-suspect area.

[0013] Furthermore, the pipeline airflow analysis module also includes a pipeline anomaly identification strategy. The pipeline anomaly identification strategy includes calculating the deviation rate between the actual flow resistance of each branch and the preset standard flow resistance based on the pipeline friction pressure loss, local pressure loss, total airflow value and total negative pressure value, and then judging the branch blockage or leakage anomaly based on the deviation rate.

[0014] Furthermore, it also includes a fire stage identification module, which constructs a fire feature vector based on smoke concentration parameters, airflow dynamic parameters, and temperature and humidity parameters collected by the aspirating detection device. The fire feature vector is then input into a pre-trained classification model to output the fire stage identification result, which includes smoldering stage, initial stage, development stage, and spread stage.

[0015] Furthermore, it also includes an airflow field update module, which releases inert gas with known release time, location and concentration at the calibration point, and collects the actual response time and response concentration of the inert gas at each sampling point, calculates the deviation of the airflow direction dynamic map, and when the deviation exceeds the preset threshold, it infers the real airflow field distribution based on the measured data, iteratively corrects the airflow direction vector and airflow velocity scalar in the airflow direction dynamic map, and generates an updated airflow direction dynamic map.

[0016] The beneficial effects of this invention are as follows: 1. By setting up a pipeline airflow analysis module, a pipeline airflow analysis model is constructed based on the fluid mechanics continuity equation, Bernoulli equation, and pipeline resistance characteristics. This model calculates in real time the airflow distribution ratio of each branch, the inlet flow rate and velocity of each sampling hole, the pressure loss along the pipeline and the local pressure loss, and the smoke particle transmission delay of each sampling branch, among other multi-dimensional parameters. This enables refined perception and dynamic inversion of the entire pipeline airflow state, providing reliable data support for the accurate determination of suspected fire areas. Furthermore, by introducing an airflow field compensation and correction module, the smoke drift of the suspected fire area output by the pipeline airflow analysis module is compensated and corrected in reverse. This effectively eliminates the disturbance effect of the fixed airflow field within the protected area on the smoke diffusion path, significantly improves the accuracy of fire point location, and achieves step-by-step focusing from regional-level location to equipment-level location.

[0017] 2. The airflow field compensation and correction module in this invention supports two correction modes. First, based on a preset dynamic airflow direction diagram, the calibration distance is calculated using the airflow direction vector and airflow velocity scalar, and the suspected fire area is shifted in the opposite direction of the airflow to obtain the corrected fire area. Second, the smoke drift distance is calculated in real time based on multi-dimensional sensing data from each sampling point, and reverse compensation correction is performed through smoke concentration gradient analysis and drift direction determination. The two modes can be selected or used in combination to adapt to the needs of different application scenarios. Among them, the correction mode based on the preset dynamic airflow direction diagram can quickly complete the positioning correction in scenarios where the airflow field in the protected area is relatively stable. It has low computational load and fast response speed, and is suitable for fire early warning scenarios with high real-time requirements. The correction mode based on real-time smoke drift distance calculation does not require the pre-construction of an airflow field model. It can independently complete the correction based solely on real-time collected multi-dimensional sensing data. It is suitable for protected areas with complex and variable airflow fields or where airflow field data cannot be obtained in advance, and has good scenario adaptability and deployment flexibility. Attached Figure Description

[0018] Figure 1 This is a flowchart of Embodiment 1 of the present invention; Figure 2 This is a flowchart of Embodiment 2 of the present invention; Figure 3 This is a flowchart of Embodiment 3 of the present invention; Figure 4 This is the light sensing curve of the photoelectric sensor in this invention; Figure 5 This is a diagram showing the dynamic airflow direction of the air sampling pipeline network in this invention. Detailed Implementation

[0019] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Identical components are denoted by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom surface," "top surface," "inner," and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0020] Example 1 This embodiment provides a fire point location system that integrates aspiration detection and pipeline airflow analysis, such as... Figure 1 As shown, the core of this system lies in correcting the smoke drift of the suspected fire area output by the pipeline airflow analysis module by using a preset dynamic airflow direction diagram.

[0021] The aspirating sensing module employs a dual-light source LED aspirating smoke detector. Its core detection principle is based on a dual-wavelength ratio algorithm. The detector contains a detection cavity housing two LED light sources and receivers of different wavelengths. When an air sample is drawn into the detection cavity, smoke particles scatter the light at different degrees across the two wavelengths. The receiver measures the intensity of the scattered light at each wavelength. Specifically, the detection cavity is a sealed optical cavity containing a light source emitter and a photoelectric receiver. The optical paths between the light source and receiver are arranged at a certain angle (usually 90° or 120°) to avoid... Direct light interference occurs when an air sample enters the detection cavity through the air inlet. Smoke particles generate scattered light when they flow through the optical path intersection area. The receiver converts the received scattered light signal into an electrical signal, which is then amplified and converted from analog to digital to obtain the scattered light intensity value. In this embodiment, two light sources are used: 570nm (blue light band) and 950nm (infrared light band). When light passes through particles, if the particle size is similar to the wavelength of the light, Mie scattering will occur. According to the Mie scattering theory, the intensity of the scattered light has a specific relationship curve with the wavelength / particle size ratio. Particles of different sizes have significantly different scattering efficiencies for different wavelengths of light.

[0022] Smoke is essentially composed of fine particulate matter. Heating and combustion tests on common insulating materials (including non-flame-retardant materials such as polyethylene, polyvinyl chloride, and cross-linked polyethylene, as well as flame-retardant materials such as flame-retardant polyethylene, flame-retardant polyvinyl chloride, and low-smoke halogen-free materials) revealed that the particle size distribution of the emitted smoke indicates that during heating and combustion, the particle size of the emitted smoke from common insulating materials is approximately 300nm-800nm. This particle size, compared to the wavelengths of 570nm and 950nm light sources, falls precisely within the sensitive band of Mie scattering. Figure 4 As shown in the figure, the photosensitive curve of the photoelectric sensor used in this embodiment shows that the 570nm wavelength has a high sensitivity response to small-diameter smoke particles with a diameter of about 300nm-500nm, which can accurately capture small-diameter smoke particles generated in the very early smoldering stage of a fire and realize early warning of a fire. The 950nm wavelength is relatively more sensitive to the scattering response of large-diameter dust, water vapor and other interfering substances. The two light sources complement each other and effectively improve the ability to distinguish particles of different diameters.

[0023] The fire stage identification module in this embodiment is based on a dual-light source ratio algorithm, which combines smoke concentration parameters, airflow dynamic parameters, and temperature and humidity parameters to construct a fire feature vector. Specifically, the feature vector includes: blue light scattering intensity. Infrared light scattering intensity Scattering ratio Smoke concentration change rate (Obtained by differential calculation of smoke concentration values ​​at continuous time points), airflow change rate ,temperature ,humidity , where the scattering ratio The key characteristic is that small-diameter smoke particles cause... High values ​​are caused by large-particle dust. The value is relatively low. This ratio system can effectively distinguish fire smoke from environmental interference, achieving a balance between high sensitivity and low false alarms. The above feature vectors are input into a pre-trained classification model. During the model training phase, at least 250 sets of measured data for each of the four stages of a fire (smoldering stage, initial stage, development stage, and spread stage) are collected (total ≥1000 sets). Professionals label the stages. Using the feature vectors as input and the fire stage as output, a random forest classification model is trained with 100 decision trees to ensure an accuracy rate ≥99.5%. During the real-time inference phase, the model outputs the fire stage and probability value corresponding to the current data. When the probability is ≥80%, it is determined to be the corresponding stage, and the corresponding warning level (warning, action, fire alarm 1, fire alarm 2) is triggered. This classification model can automatically output the corresponding warning level according to the fire development stage, providing a basis for subsequent differentiated linkage strategies.

[0024] The hardware of the fire location system integrating aspirating detection and pipeline airflow analysis also includes an industrial control computer, an aspirating sensing module, an air sampling pipeline network, and a fire location module. The air sampling pipeline network includes, for example,... Figure 5 As shown, it includes several sampling branches and sampling holes arranged in the protected area. The sampling holes can be the air inlets of the aspirating detector. The length, diameter and spatial position of each branch pipe are preset according to the pipe network topology of the protected area.

[0025] Based on the aforementioned hardware setup, the ignition point location system integrating aspirating detection and pipeline airflow analysis first pre-constructs a pipeline airflow analysis model. This model establishes a mathematical framework based on the fluid mechanics continuity equation, Bernoulli's equation, and pipeline resistance characteristics. The formula for calculating the friction resistance along the pipeline is as follows: ,in, This is the friction factor, which is related to the pipe roughness and Reynolds number. As the head of the department, For pipe diameter, This is the air density (corrected for temperature compensation). The average airflow velocity inside the pipe; the formula for calculating the local resistance of the pipe is: ,in, The local resistance coefficient was determined experimentally for local components such as sampling holes, elbows, and tees; the total flow distribution model... Based on the principle of flow resistance balance in each branch, the flow rate in each branch is determined by the balance of branch resistance. ,in, For the first airflow rate of the branch, For the first The total flow resistance of the branch (including frictional flow resistance and local flow resistance, the formula for calculating the total flow resistance is as follows) , (where the branch cross-sectional area is used), the formula for calculating the smoke particle transmission delay model is: ,in, This is the time required for smoke particles to travel from the sampling port to the detection unit. As the branch road manager, The average airflow velocity within the pipe is used. The physical meaning of this time delay model is that smoke particles are transported with the airflow within the sampling pipe, and their transport velocity is approximately equal to the average airflow velocity within the pipe. Therefore, the transport delay is equal to the pipe length divided by the flow velocity. This time delay model provides a quantitative basis for determining the branch from which the smoke originates. The branch with the shortest time delay is the branch closest to the fire source. Based on the actual pipe network topology of the protected area, the number of branches, the length of each branch, the diameter of each branch, the local resistance coefficient of each branch, the number of each sampling hole, and the spatial location of each sampling hole are input into the above mathematical framework to generate a pipe network airflow analysis model that matches the current protected area. The model constructed based on fluid dynamics equations can accurately calculate the airflow distribution ratio and time delay parameters of each branch, providing a reliable data foundation for subsequent area determination.

[0026] When an ignition point is detected and smoke is generated, the system begins operation. The aspirating sensing module collects multi-dimensional sensing data in real time through the air sampling pipeline network, including at least smoke concentration parameters and airflow dynamic parameters. The pipeline airflow analysis module then solves the multi-dimensional sensing data using a preset pipeline airflow analysis model. Specifically, firstly, based on the real-time collected total airflow value and total negative pressure value at the detection host, the system calculates the total airflow value (measured by the flow sensor built into the host) and the total negative pressure value (measured by the negative pressure sensor built into the host). Both are analog signals, which are converted to digital values ​​for calculation. The system calculates the airflow distribution ratio of each branch, the inlet flow rate and velocity of each sampling hole, the pressure loss along the pipeline and the local pressure loss by combining the preset pipeline topology with the total flow distribution model and the flow resistance balance equation of each branch. The inlet flow rate of each sampling hole is approximately estimated by dividing the flow rate of each branch by the number of sampling holes on that branch, or by finely distributing it according to the difference in the local resistance coefficient of each sampling hole. By monitoring the total airflow value and the total negative pressure value in real time and dynamically reversing the state of the entire pipeline network, the system can detect pipeline anomalies in a timely manner and avoid positioning failures caused by pipeline faults.

[0027] Secondly, based on the length of each branch pipe and the average airflow velocity within the pipe, the smoke particle transmission delay of each sampling branch is calculated using a smoke particle transmission delay model. The strategy for determining suspected fire areas executes the following steps: Invalid sampling branches are eliminated based on the airflow rate and velocity of each sampling hole. When the airflow rate of a sampling hole is lower than a certain percentage of the normal flow rate, it is determined that the sampling hole may be blocked, and the sampling branch is marked as invalid and eliminated. Among the remaining valid sampling branches, candidate smoke source branches are determined based on the smoke particle transmission delay. The transmission delays of each branch are compared, and the sampling branch with the smallest transmission delay is determined as the candidate smoke source branch. The source branch of the smoke is identified, and the confidence level is verified based on the airflow distribution ratio of the candidate smoke source branch. If the airflow distribution ratio of the branch is consistent with the trend of smoke concentration change, such as the branch with a high airflow distribution ratio corresponding to an increase in smoke concentration, then the judgment result is confirmed to have high confidence. Through a three-layer judgment mechanism of data validity screening to eliminate blocked branches, time delay comparison to determine candidate source branches, and airflow distribution ratio verification of confidence, the interference of invalid data and low confidence results is effectively eliminated, and the reliability of area judgment is improved. Finally, the spatial coverage area corresponding to the verified candidate smoke source branches is determined as the suspected fire area.

[0028] A dynamic airflow pattern is pre-constructed, obtained through fluid dynamics simulation of the spatial geometry and ventilation layout of the protected area, such as... Figure 5 As shown, the simulation includes airflow direction vectors and airflow velocity scalars at various spatial locations within the protected area. The simulation process involves establishing a three-dimensional geometric model of the protected area, setting boundary conditions including the location of vents, supply air velocity, and return air vent location, dividing the computational grid, and using RANS equations or LES models to solve the steady-state or transient flow field. After convergence, the airflow direction vectors and velocity scalars of each grid node are extracted, and the data at the sampling hole locations are interpolated. For example, in the protected area of ​​a cabinet room, the air conditioning vent is located on the north wall. The simulation calculation shows that the airflow direction in this area is from north to south, and the airflow velocity scalar decreases with distance, reaching approximately 0.5 m / s at 5 meters from the vent. After actual deployment, the simulation results can be corrected by conducting on-site measurements at each sampling point to further improve the matching degree between the dynamic diagram and the actual working conditions.

[0029] The determination of the estimated residence time of smoke particles in the airflow field specifically includes: First, determining the spatial location of each sampling point within the suspected fire area, and associating and matching each sampling point with the corresponding position of the airflow direction vector and airflow velocity scalar in the dynamic airflow diagram, i.e., finding the airflow parameters of the grid node closest to the sampling point coordinates in the dynamic diagram; Second, obtaining the actual arrival time of smoke particles from the fire source to each sampling point, i.e., extracting the time point when the concentration first exceeds the warning threshold from the smoke concentration time series of the dual-source detector. This time has deducted the transmission delay of smoke particles in the pipeline network. The method for extracting the actual arrival time is to perform low-pass filtering and noise reduction on the smoke concentration time series, and then calculate the concentration difference between adjacent time points. When the difference value continuously exceeds the preset threshold, the actual arrival time is determined. The arrival time of the smoke is determined by a threshold. Then, the theoretical transmission time is calculated based on the spatial distance between each sampling point and the center of the suspected fire area, as well as the airflow velocity scalar at the corresponding location. Theoretical transmission time = spatial distance / airflow velocity scalar. Finally, the difference between the actual arrival time and the theoretical transmission time is used as the estimated residence time of the smoke particles in the airflow field. When the difference is positive, it indicates that the smoke is pushed by the airflow, and the actual arrival time is later than the theoretical value. When the difference is negative, it indicates that the smoke is propagating against the airflow, and the actual arrival time is earlier than the theoretical value. In this case, the absolute value is taken as the residence time. This difference quantifies the degree of influence of airflow disturbance on smoke drift, providing a quantitative basis for subsequent translation correction. The larger the difference, the more significant the airflow disturbance, and the greater the distance that needs to be corrected.

[0030] The formula for calculating the calibration distance is: ,in, To calibrate the distance, For airflow velocity scalar, To estimate the residence time, assuming a fire scenario, the suspected fire area is located 0.5 meters east of cabinet A, with airflow direction from east to west and airflow speed of 0.5 m / s, and an estimated residence time of 2 seconds (meaning it takes 2 seconds for smoke to drift from the actual fire source to the sampling point, while theoretically it would only take a short time without airflow disturbance; this difference indicates that the smoke is deflected by the airflow), the calibration distance is 1 meter west. The airflow field compensation and correction module translates the suspected fire area 1 meter in the opposite direction of the airflow, resulting in a corrected fire area 0.5 meters west of cabinet A. The translation operation updates the coordinates of the boundary points of the suspected fire area one by one to the original coordinates minus the calibration distance multiplied by the direction unit vector, keeping the area shape unchanged. By translating and correcting in the opposite direction of the airflow, the disturbance effect of the fixed airflow field on the smoke diffusion path is effectively eliminated, making the corrected fire area closer to the actual fire source location.

[0031] The precise fire location module, based on the corrected fire area, calls upon the detection data from the location sub-modules deployed in each device within that area. Each location sub-module is deployed inside or on top of a cabinet, power distribution cabinet, or precision equipment, and possesses independent smoke concentration detection capabilities and communication address codes. When the corrected fire area covers cabinet A, the precise fire location module retrieves the smoke concentration parameters fed back by the location sub-modules inside and around cabinet A and adjacent cabinets. A weighted fusion algorithm is used for fusion judgment, assigning weights based on the spatial distance between each location sub-module and the center of the corrected fire area. Location sub-modules that are closer are given higher weights. Finally, the precise coordinate information of the fire point is output. Through the cabinet-level deployment and weighted fusion judgment of the location sub-modules, a step-by-step focusing from area-level positioning to device-level positioning is achieved. The final output fire point coordinates can be accurate to the specific device. The output format is device name, cabinet level, and specific location description, such as "cabinet A - 3rd layer - left-side device".

[0032] It also includes a pipeline anomaly identification strategy, which calculates the deviation rate between the actual flow resistance of each branch and the preset standard flow resistance based on the pressure loss along the pipeline and the local pressure loss, combined with the total airflow value and total negative pressure value collected in real time at the detection host. When the deviation rate exceeds a preset threshold, the branch is determined to be blocked (manifested as decreased airflow, increased negative pressure, and abnormally low flow rate); when the deviation rate is below a preset threshold, the branch is determined to be leaking (manifested as excessively high airflow, abnormally low negative pressure, and abnormally excessive flow rate). The system outputs diagnostic results including the type of abnormality and the location information of the abnormal branch, pushes fault information and maintenance suggestions to the industrial control computer interface, and prompts maintenance personnel to check the abnormal branch. By calculating the deviation rate based on resistance parameters, the system can actively detect pipeline blockage or leakage and accurately locate the abnormal branch, effectively reducing the risk of location failure caused by pipeline faults.

[0033] Example 2 Based on Example 1, the correction mode of the airflow field compensation and correction module is replaced with reverse compensation and correction performed by real-time calculation of the smoke drift distance based on multi-dimensional sensing data from each sampling point. Figure 2 As shown, firstly, the smoke concentration parameters of each sampling point within the suspected fire area are obtained within the same time window. The smoke concentration parameters include the dual-channel scattering intensities at 570nm and 950nm and their ratio. Due to the ratio of dual light sources The influence of environmental contaminants has been eliminated. This embodiment uses only the effective smoke concentration data verified by the ratio for gradient analysis. Through dual-light source ratio preprocessing, the contamination of concentration data by environmental contaminants such as dust and water vapor is eliminated, making the gradient analysis more accurate. The smoke concentration difference between adjacent sampling points is calculated to construct the spatial gradient distribution field of smoke concentration. The gradient is calculated numerically as follows: for sampling points The concentration values ​​of the surrounding adjacent sampling points are taken, and the differences in each direction are calculated. The direction of the maximum difference is taken as the gradient direction, and the difference value is taken as the gradient amplitude. For example, in a certain protected area, there are 5 sampling points. The effective smoke concentration values ​​after dual light source ratio verification are: sampling point A (120ppm), sampling point B (100ppm), sampling point C (80ppm), sampling point D (60ppm), and sampling point E (40ppm). The concentration difference between adjacent sampling points shows a gradient distribution that decreases from A to E. This gradient distribution field intuitively reflects the diffusion direction and attenuation trend of smoke in space.

[0034] Secondly, based on the direction of the fastest decrease in concentration value in the smoke concentration gradient distribution field (i.e. The direction of the smoke particles is determined by combining the arrival time sequence of the smoke particles at each sampling point (the timestamp of the first appearance of the smoke is extracted from the raw data of the dual-source detector, after deducting the pipeline transmission delay). The fusion rule for direction determination is as follows: when the gradient direction is consistent with the time sequence direction, the direction is taken as the drift direction with higher confidence; when the two are inconsistent, the average of the angle between the two is calculated, and the time sequence direction is given higher weight. In the example above, the direction with the fastest decrease in concentration value is from A to E, and the arrival time sequence of the smoke particles is A earliest and E latest. Since the two are consistent, the actual drift direction is determined to be from A to E. If the gradient direction is inconsistent with the arrival time sequence, that is, the concentration gradient decreases from A to E, but the arrival time sequence is B earliest, then the arrival time sequence is the main basis, because the time sequence directly reflects the order of smoke arrival and is more reliable.

[0035] Then, based on the actual drift direction and the spatial distance between each sampling point, the distance from the smoke particle drifting from the fire source to the earliest sampling point where the smoke was captured is calculated as the smoke drift distance. Specifically, taking sampling point A, which has the highest concentration and the earliest arrival time, as the reference point, the location of the fire source is estimated along the opposite direction of the drift direction, i.e., the direction of the smoke source, based on the concentration decay gradient. The estimation method is as follows: extrapolate the concentration value point by point along the opposite direction of the drift with a step size of 0.1 meters. The extrapolation model can use linear decay. ,in To be at a distance from the fire source Smoke concentration at the location The actual smoke concentration at the sampling point. For distance variable, attenuation coefficient The smoke drift distance is calculated by dividing the concentration difference between adjacent sampling points by the distance. When the extrapolated concentration value drops to the background concentration (the background concentration value under normal conditions), the corresponding distance is the smoke drift distance. Assuming that the distance between adjacent sampling points is 5 meters along the drift direction, and the concentration decreases approximately linearly from A to E (attenuation rate of about 20 ppm / m), then the fire source may be located about 2.5 meters west of sampling point A. The concentration is linearly extrapolated from 120 ppm to 0 ppm, and the smoke drift distance is 2.5 meters. To improve the estimation accuracy, an exponential decay model can be used. Perform fitting, where The source of the fire ( The initial smoke concentration at (=0), The attenuation coefficient is used to estimate the location of the fire source by using the concentration attenuation gradient. The drift distance can be calculated independently without the need for a preset airflow field model, providing a basis for reverse correction.

[0036] The airflow field compensation and correction module, based on the calculated smoke drift distance, translates the suspected fire area by the distance in the opposite direction of the drift direction, and outputs the corrected fire area. The specific implementation of the translation operation is the same as in Example 1. The coordinates of the boundary points of the area are translated one by one in the opposite direction of the drift, and the translation distance is the smoke drift distance. When the suspected fire area is irregular in shape, the geometric similarity of the shape is maintained, and only the position of its geometric center and boundary points is moved. Assuming that the suspected fire area was originally located within 3 meters around sampling point A, and the drift direction is from west to east as determined by the pipeline airflow analysis module based on the time delay inversion, and the smoke drift distance is 2.5 meters, then the corrected fire area is located 2.5 meters west of sampling point A, that is, the original suspected fire area is translated 2.5 meters west.

[0037] In addition, a combination of two correction modes is used: a dynamic airflow map and a reverse compensation correction based on real-time calculation of smoke drift distance using multi-dimensional sensing data from each sampling point. The switching logic for this combination is as follows: During system initialization, the system defaults to parallel operation of both modes, comparing the correction results output by each. When the deviation between the correction results from both modes is less than a preset threshold multiple times consecutively, it indicates that the dynamic map is highly accurate, and the system can switch to a single correction mode based on the dynamic map to reduce computational costs. When the deviation between the real-time calculation result and the dynamic map correction result exceeds a preset range, the system automatically switches back to the parallel dual-mode state, using the real-time calculation result as the primary correction basis, and simultaneously triggering the dynamic map update process. Specifically... In the initial stage of system deployment, the preset airflow direction dynamic map may not be accurate enough due to simulation errors or changes in field conditions. At this time, the real-time calculated smoke drift distance is used as the main correction basis. As the system runs for a longer period of time, the preset dynamic map is corrected by calculating multiple times. That is, the drift distance and direction calculated multiple times are statistically averaged and fed back to correct the airflow vector in the dynamic map, gradually improving its accuracy. After the accuracy of the dynamic map is verified to meet the standard, it can be switched to a fast correction mode based on the dynamic map to reduce the computational cost. The two modes can be automatically switched according to the confidence threshold: when the deviation between the real-time calculation result and the dynamic map correction result exceeds the preset range, the real-time calculation result is used first and the dynamic map update process is triggered.

[0038] Example 3 Based on Example 1, this embodiment adds an airflow field update module for actively verifying and closed-loop correcting the accuracy of the dynamic airflow direction diagram. This added airflow field update module is connected to the airflow field compensation and correction module to verify and adjust the accuracy of the dynamic airflow direction diagram. This module includes a tracer gas release device, a data acquisition and deviation calculation unit, and a dynamic diagram correction unit. The interface protocol between the airflow field update module and the airflow field compensation and correction module is as follows: the update module reads the dynamic diagram file currently used by the compensation and correction module, generates a corrected dynamic diagram file, and writes it back to the storage area of ​​the compensation and correction module, replacing the original file. By adding the airflow field update module, the system possesses adaptive capabilities, enabling it to cope with long-term drift of the airflow field caused by changes in equipment operating status, environmental conditions, and other factors.

[0039] The tracer gas release device includes several inert gas releasers deployed at preset calibration points within the protected area. Each releaser has an independent control valve and communication address code, and can release inert gas according to a preset timing sequence or remote command. The releasers should be placed near ventilation openings, airflow intersections, or sparse sampling areas. In this embodiment, carbon dioxide is used as the tracer gas because it is non-flammable, does not react with the environment, and the aspirating sensing module of this system has special gas concentration detection capabilities, so there is no need to add an additional gas sensor. The spatial location, release time, and release concentration of each calibration point are pre-recorded in the system.

[0040] Specifically, such as Figure 3 As shown, firstly, the tracer gas release strategy is executed. The system selects one or more preset verification points and releases them one by one to avoid concentration superposition interference. The corresponding release device is controlled to release a quantitative amount of inert gas, and the precise time, release location and release concentration of each release are recorded. The one-by-one release strategy avoids the concentration superposition interference of multi-source gas release, makes the response data of each verification point independent of each other, and improves the reliability of the verification results.

[0041] Secondly, the actual data acquisition strategy is implemented. The inhalation sensing module collects the actual response time and response concentration of inert gas at each sampling point. The actual response time is the time interval from the moment the gas is released to the moment the carbon dioxide concentration at the sampling point first exceeds a certain threshold of the background concentration. The actual response concentration is defined as the peak carbon dioxide concentration detected at the sampling point after the gas is released. Since the system has dual-light source detection capability, it can simultaneously collect carbon dioxide concentration for verification. The system compares the collected actual response time with the theoretical response time calculated based on the current airflow direction dynamic diagram. The theoretical response time includes two parts: the air propagation delay of the gas from the release point to the sampling point (calculated based on the airflow speed and spatial distance in the dynamic diagram; specifically, the propagation path is obtained by integrating along the airflow streamline from the release point to the sampling point, and the path length is divided by the local airflow speed to obtain the propagation delay) plus the pipeline transmission delay from the sampling point to the host via the pipeline network (calculated by the smoke particle transmission delay model). At the same time, the actual response concentration is compared with the theoretical diffusion concentration calculated based on the dynamic diagram, and the time deviation and concentration deviation are calculated respectively.

[0042] Combining both, the overall deviation of the airflow direction dynamic diagram under the current operating conditions is obtained. The formula for calculating the deviation is: ,in, , For preset weighting coefficients, and These are the actual and theoretical response times, respectively. and These represent the actual and theoretical response concentrations, respectively. When the deviation is 0, it indicates that the dynamic graph is completely accurate. The larger the deviation, the greater the deviation between the dynamic graph and the actual airflow field. Through the quantitative evaluation of the overall deviation, the system can clearly determine whether the dynamic graph needs to be corrected. The higher the deviation, the greater the deviation between the current dynamic graph and the actual airflow field. When the deviation exceeds the preset accuracy threshold, the dynamic graph correction process is triggered.

[0043] Finally, a dynamic graph correction strategy is implemented. The system infers the true airflow field distribution based on the measured inert gas response data at each sampling point. The specific inference method is as follows: taking the release location of the verification point as the source point and the actual response time and response concentration of each sampling point as constraints, an inverse optimization problem is established. The objective function is to minimize the weighted square error between the theoretical response value calculated based on the corrected dynamic graph and the measured value. Gradient descent or particle swarm optimization algorithms are used. In each iteration, the airflow direction vector and airflow velocity scalar in the dynamic graph are gradually adjusted along the gradient direction until the deviation between the theoretical response value calculated based on the corrected dynamic graph and the measured value converges to within the threshold. After generating the updated airflow direction dynamic graph, the data integrity is ensured by hash verification, and then it is stored in the airflow field compensation and correction module to replace the original dynamic graph. An update log is recorded, including the update time, update reason, and deviation change, for subsequent correction of suspected fire areas.

[0044] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A fire ignition point location system integrating aspiration detection and pipeline airflow analysis, characterized in that: include: The aspirating sensing module collects multi-dimensional sensing data in real time, including at least smoke concentration parameters and airflow dynamic parameters, through aspirating detection devices deployed in the air sampling pipeline network. The pipeline airflow analysis module calculates the multi-dimensional sensing data through a preset pipeline airflow analysis model, generating airflow distribution ratios, smoke particle transmission delays, and pipeline resistance characteristic parameters for each sampling branch. Based on the calculation results, it inverts the smoke source area and outputs the suspected fire area. The airflow field compensation and correction module performs reverse compensation and correction of smoke drift in the suspected fire area based on the preset dynamic airflow direction diagram and / or the smoke drift distance calculated from the multi-dimensional sensing data of each sampling point, and outputs the corrected fire area. The precise fire point location module calls upon the detection data from the location sub-modules deployed in various devices within the fire area, fuses and judges the smoke concentration parameters fed back by the location sub-modules, and outputs the precise coordinate information of the fire point.

2. The system of claim 1, wherein: The dynamic airflow diagram contains airflow direction vectors and airflow velocity scalars at various locations in space. The calibration distance is calculated based on the airflow velocity scalar and the estimated residence time of smoke particles in the airflow field. The suspected fire area is shifted along the opposite direction of the airflow by the calibration distance to obtain the corrected fire area.

3. The system of claim 2, wherein: The airflow compensation and correction module includes an estimated residence time calculation strategy. The estimated residence time calculation strategy includes determining the spatial location of each sampling point in the suspected fire area, associating and matching each sampling point with the airflow direction vector and airflow velocity scalar at the corresponding position in the airflow direction dynamic diagram, obtaining the actual arrival time of smoke particles from the fire source to each sampling point, calculating the theoretical transmission time based on the spatial distance between each sampling point and the center position of the suspected fire area and the airflow velocity scalar at the corresponding position, and using the difference between the actual arrival time and the theoretical transmission time as the estimated residence time of smoke particles in the airflow field.

4. The system of claim 1, wherein: The airflow compensation and correction module includes a smoke drift distance calculation strategy. This strategy involves obtaining the smoke concentration parameters of each sampling point within the suspected fire area within the same time window, calculating the smoke concentration difference between adjacent sampling points, and constructing a spatial gradient distribution field of smoke concentration. Based on the direction of the fastest decrease in concentration value in the smoke concentration gradient distribution field, and combined with the arrival time sequence of smoke particles at each sampling point, the actual drift direction of the smoke particles is determined. Then, based on the actual drift direction and the spatial distance between each sampling point, the distance from the fire source to the sampling point where the smoke was first captured is calculated as the smoke drift distance.

5. The system according to any one of claims 1-4, wherein: The pipeline airflow analysis module includes a pipeline airflow analysis model construction strategy. The pipeline airflow analysis model construction strategy includes a mathematical framework for constructing a pipeline airflow analysis model based on the fluid mechanics continuity equation, Bernoulli equation and pipeline resistance characteristics. The pipeline resistance characteristics include the friction resistance calculated based on the friction coefficient, pipe length, pipe diameter and average airflow velocity in the pipe, and the local resistance calculated based on the local resistance coefficient and average airflow velocity in the pipe. Based on the preset pipeline topology, the number of branches, the length of each branch, the diameter of each branch, the local resistance coefficient of each branch, the number of each sampling hole, and the spatial location of each sampling hole are input into the mathematical framework to generate a pipeline airflow analysis model that includes a total flow distribution model and a smoke particle transmission delay model. The total flow distribution model is used to calculate the airflow distribution ratio of each branch and the inlet flow rate and velocity of each sampling hole based on the total flow resistance of each branch. The smoke particle transmission delay model is used to calculate the transmission delay required for smoke particles to travel from the sampling hole to the detection host.

6. The system of claim 5, wherein: The pipeline airflow analysis module also includes a real-time pipeline status calculation strategy. This strategy involves calculating the airflow distribution ratio of each branch, the inlet flow rate and velocity of each sampling hole, the pressure loss along the pipeline, and the local pressure loss based on the total airflow value, total negative pressure value, and pipeline topology acquired by the real-time acquisition and detection host, using a total flow distribution model and the flow resistance balance equation of each branch. Based on the pipe length and the average airflow velocity in each branch, the smoke particle transmission delay of each sampling branch is calculated using a smoke particle transmission delay model.

7. The system of claim 6, wherein: The pipeline airflow analysis module also includes a fire-suspect area determination strategy. The fire-suspect area determination strategy includes eliminating invalid sampling branches based on the airflow and velocity of each sampling hole, determining candidate smoke source branches based on the smoke particle transmission delay among the remaining valid sampling branches, verifying high confidence based on the airflow distribution ratio of the candidate smoke source branches, and determining the spatial coverage area corresponding to the verified candidate smoke source branches as the fire-suspect area.

8. The system of claim 6, wherein: The pipeline airflow analysis module also includes a pipeline anomaly identification strategy. The pipeline anomaly identification strategy includes calculating the deviation rate between the actual flow resistance of each branch and the preset standard flow resistance based on the pressure loss along the pipeline, the local pressure loss, the total airflow value and the total negative pressure value, and then judging the branch blockage or leakage anomaly based on the deviation rate.

9. The system of claim 1, wherein: the system is configured to determine a location of a fire source based on the analysis of the gas flow in the pipe network. It also includes a fire stage identification module, which constructs a fire feature vector based on smoke concentration parameters, airflow dynamic parameters, and temperature and humidity parameters collected by the aspirating detection device. The fire feature vector is then input into a pre-trained classification model to output the fire stage identification result. The fire stages include smoldering stage, initial stage, development stage, and spread stage.

10. The ignition point location system integrating aspiration detection and pipeline airflow analysis according to claim 3, characterized in that: It also includes an airflow field update module, which releases inert gas with known release time, location and concentration at the calibration point, and collects the actual response time and response concentration of the inert gas at each sampling point. It calculates the deviation of the airflow direction dynamic map. When the deviation exceeds the preset threshold, it infers the real airflow field distribution based on the measured data, and iteratively corrects the airflow direction vector and airflow velocity scalar in the airflow direction dynamic map to generate an updated airflow direction dynamic map.

Images