Fire point location precision tracking method based on distributed optical fiber

Abstract

This invention discloses a method for precise tracking of fire ignition points based on distributed optical fibers, specifically relating to the field of fire monitoring and safety early warning technology. The method involves acquiring the initial temperature distribution sequence along the optical fiber and periodically collecting temperature data during monitoring. Temperature anomaly segments are identified based on the temperature increment distribution, and the temperature rise gradient characteristics and spatial diffusion width characteristics of these anomaly segments are extracted. The asymmetric temperature rise gradient coefficient and thermal diffusion offset index are calculated to construct an anomaly feature set. A thermal diffusion evolution model is constructed based on this feature set, and reverse heat source fitting calculations are performed on the temperature anomaly segments to obtain a set of candidate ignition point locations. Trajectory correlation analysis is conducted on the candidate ignition point locations at multiple consecutive time points, and the point with the smallest spatial drift and continuously increasing temperature rise is selected as the target ignition point. This invention can effectively suppress positioning deviations caused by thermal diffusion and environmental disturbances, achieving precise identification and dynamic tracking of early fire source locations in complex environments.

Description

Technical Field

[0001] This invention relates to the field of fire monitoring and safety early warning technology, specifically to a method for precise tracking of fire ignition points based on distributed optical fibers. Background Technology

[0002] In narrow, enclosed spaces such as large underground utility tunnels, data center cable trays, and sealed energy storage chambers, cable overload or localized insulation aging often triggers microscale smoldering fires. These fires initially manifest as localized temperature rises on the order of centimeters, accompanied by slow air convection, causing heat to diffuse along the structural surface. This makes it difficult for conventional point-based temperature or smoke sensors to pinpoint the actual fire location in a timely manner. While distributed fiber optic temperature measurement technology can acquire continuous temperature information along the line, it is susceptible to factors such as equipment heat dissipation, localized ventilation disturbances, and delayed heat diffusion in complex environments, resulting in temperature rise peak drift or multiple peak superposition. This leads to accumulated fire source location errors, making it difficult to accurately track early, concealed fire points. Therefore, how to dynamically analyze distributed fiber optic temperature data under complex thermal disturbance environments to accurately determine and continuously track the actual fire location has become an urgent technical problem to be solved. Summary of the Invention

[0003] The purpose of this invention is to provide a method for accurately tracking the location of fire ignition points based on distributed optical fibers, in order to address the shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for precise tracking of fire ignition points based on distributed optical fibers, comprising:

[0005] Distributed optical fiber sensing lines are laid along the target monitoring area to obtain the corresponding initial temperature distribution sequence T0(x), where x is the spatial position along the optical fiber line;

[0006] During the monitoring process, temperature data along the optical fiber is periodically collected to obtain the temperature distribution sequence Tt(x) at different times, and the temperature increment distribution ΔTt(x) between adjacent times is calculated.

[0007] Based on ΔTt(x), temperature anomaly segments are identified, and the temperature rise gradient characteristics and spatial diffusion width characteristics of the anomaly segments are extracted. Asymmetric temperature rise gradient coefficient Kg and thermal diffusion offset index Kd are generated to form an anomaly feature set B=(Kg,Kd).

[0008] A thermal diffusion evolution model is constructed based on the abnormal feature set B, and the reverse heat source fitting calculation is performed on the abnormal section to obtain the set of candidate ignition point locations.

[0009] Trajectory correlation analysis is performed on the set of candidate ignition point locations obtained at multiple consecutive time points to select the point with the smallest spatial drift and continuous temperature increase as the target ignition point, and the precise location and dynamic trajectory of the target ignition point are output.

[0010] Preferably, spatial position coordinates are established along the length of the optical fiber, the starting end of the optical fiber is defined as the spatial starting point, a continuous spatial coordinate sequence is established along the length of the optical fiber, the optical fiber is divided into multiple discrete spatial sampling units, each spatial sampling unit corresponds to a temperature measurement position, for the i-th spatial sampling unit along the optical fiber, the initial temperature value is calculated, and the initial temperature values ​​of each spatial sampling unit are arranged according to the spatial position order of the optical fiber to construct the initial temperature distribution sequence T0(x), where x is the spatial position along the optical fiber.

[0011] Preferably, the temperature increment distribution ΔTt(x) between adjacent time points is the temperature change at position x along the fiber between time t and time t-1.

[0012] Preferably, identifying temperature anomaly zones based on temperature increment distribution includes the following steps:

[0013] A continuous temperature increment curve is constructed based on the temperature change values ​​at each spatial location in the temperature increment distribution. The continuous temperature increment curve is then smoothed using a sliding window. A smooth temperature increment sequence is formed by taking the arithmetic mean of the temperature increments within the window.

[0014] In the smoothed temperature increment sequence, the temperature increment gradient between adjacent spatial sampling units is calculated, and the positions where the temperature increment gradient is greater than a preset gradient threshold are marked as candidate anomaly positions.

[0015] The search is gradually expanded to both sides of the candidate anomaly location. When the temperature increment of the continuous spatial sampling units is greater than the temperature increment reference threshold, the continuous interval is determined as the temperature anomaly segment.

[0016] Preferably, the calculation of the asymmetric temperature gradient coefficient includes the following steps:

[0017] The temperature increment sequence within the temperature anomaly zone is processed by empirical mode decomposition according to spatial location order. Multiple intrinsic mode components are separated by successively screening local extreme points and constructing envelope curves. The low-frequency intrinsic mode component containing the main temperature rise trend is selected as the temperature rise trend sequence.

[0018] Using the temperature peak position of the temperature rise trend sequence as the center, the temperature rise trend sequence is divided into a left temperature rise sequence and a right temperature rise sequence, and Hilbert transform is performed on the left temperature rise sequence and the right temperature rise sequence respectively to obtain the corresponding instantaneous amplitude curves.

[0019] The spatial gradient values ​​of the left and right temperature rise sequences are calculated based on the instantaneous amplitude curves, and the left and right temperature rise gradients are obtained respectively.

[0020] The asymmetric temperature rise gradient coefficient is constructed based on the ratio of the difference between the temperature rise gradient on the left and the temperature rise gradient on the right to the sum of the two gradients.

[0021] Preferably, the calculation of the thermal diffusion offset index includes the following steps:

[0022] Within the temperature anomaly zone, the temperature increment values ​​corresponding to each spatial sampling unit are extracted and normalized according to the spatial location order to convert the temperature increment values ​​into a probability density distribution, thereby constructing the actual temperature diffusion distribution function.

[0023] Using the temperature peak position of the temperature anomaly section as the center of symmetry, a reference temperature diffusion distribution function symmetrical about the temperature peak position is constructed based on the heat conduction and diffusion law.

[0024] Based on the spatial difference between the actual temperature diffusion distribution function and the reference temperature diffusion distribution function, the optimal transmission cost between the two is calculated, and the Wasserstein distance value is obtained by solving for the minimum transmission distance.

[0025] The Wasserstein distance value and the spatial diffusion width of the temperature anomaly zone are normalized to obtain the thermal diffusion offset index, which characterizes the degree of spatial offset of thermal diffusion.

[0026] Preferably, the thermal diffusion evolution model is constructed based on the abnormal feature set B, including the following steps: based on the numerical relationship between the asymmetric temperature rise gradient coefficient and the thermal diffusion migration index in the abnormal feature set, a temperature spatial gradient field is established, and the thermal diffusion evolution equation is constructed using the temperature increment value of each spatial sampling unit in the temperature anomaly segment as the initial condition, wherein the thermal diffusion evolution equation is obtained by discretization calculation of the one-dimensional heat conduction partial differential equation.

[0027] Preferably, an asymmetric temperature rise gradient coefficient is introduced as a gradient correction factor in the thermal diffusion evolution equation, and a thermal diffusion offset index is introduced as a spatial drift constraint parameter. The temperature field is iteratively evolved to obtain the predicted temperature diffusion distribution at multiple times. By minimizing the error between the actual temperature diffusion distribution and the predicted temperature diffusion distribution, the least squares inversion method is used to progressively search for possible heat source locations, resulting in several candidate heat source locations that meet the error threshold condition. The candidate heat source locations are then aggregated according to their spatial location to form a set of candidate ignition point locations.

[0028] Preferably, trajectory correlation analysis is performed on the set of candidate ignition point locations obtained at multiple consecutive time points, including the following steps:

[0029] Spatial location matching is performed on the set of candidate ignition point locations obtained from multiple consecutive time moments in chronological order. A candidate location association matrix is ​​constructed by calculating the spatial distance between candidate ignition points at adjacent time moments, and a temporal association relationship is established for candidate locations whose spatial distance is less than a preset association distance threshold. Based on the temporal association relationship, the trajectories of candidate ignition points at consecutive time moments are spliced ​​to generate multiple candidate ignition trajectories. The spatial location change amount at adjacent time moments is calculated for each candidate ignition trajectory to obtain the corresponding spatial drift amount sequence. While calculating the spatial drift amount sequence, the temperature increment change sequence at the corresponding spatial location is extracted, and the continuous growth rate of the temperature increment change sequence is used to determine the continuous temperature rise characteristic.

[0030] Preferably, by comprehensively comparing the average spatial drift and continuous temperature rise rate of each candidate ignition trajectory, the candidate ignition trajectory with the smallest average spatial drift and continuous temperature rise is selected, and the spatial location corresponding to the candidate ignition trajectory is determined as the target ignition point location. At the same time, the dynamic change trajectory of the target ignition point over time is output.

[0031] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0032] 1. This invention identifies anomalous sections in the temperature increment distribution acquired by distributed optical fibers, and further extracts the asymmetric temperature rise gradient coefficient and thermal diffusion shift index to construct an anomalous feature set. Based on this, a thermal diffusion evolution model is established and reverse heat source fitting calculations are performed to obtain a set of candidate ignition point locations. This technical solution can comprehensively characterize anomalous temperature rise behavior from two dimensions: the spatial distribution pattern of temperature and the thermal diffusion shift characteristics. It effectively overcomes the problem of temperature peak drift that easily occurs in traditional temperature peak location methods under complex thermal disturbance environments, thus significantly improving the accuracy of early fire source location. It is particularly suitable for identifying concealed initial fire sources in narrow, enclosed spaces such as cable tunnels, integrated pipe corridors, and energy storage equipment cabins.

[0033] 2. This invention performs trajectory correlation analysis on a set of candidate ignition point locations across multiple consecutive time points. By combining spatial drift and continuous temperature increase characteristics, it filters for true ignition trajectories. This ensures that the determination of the target ignition point relies not only on the temperature distribution characteristics at a single moment but also on a comprehensive judgment based on the evolutionary patterns over time. This effectively eliminates misjudgments caused by environmental heat source disturbances or intermittent heat dissipation from equipment. By outputting the precise spatial location and dynamic trajectory of the target ignition point, continuous tracking of the fire source development process can be achieved, providing more reliable data support for fire early warning and rapid response, and significantly improving the stability and reliability of early fire monitoring. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0035] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] For examples, please refer to Figure 1 As shown in this embodiment, the method for precise tracking of fire ignition points based on distributed optical fiber includes:

[0038] Distributed optical fiber sensing lines are laid along the target monitoring area to obtain the corresponding initial temperature distribution sequence T0(x), where x is the spatial position along the optical fiber line.

[0039] In this embodiment, a distributed optical fiber sensing line is laid along the target monitoring area to acquire the corresponding initial temperature distribution sequence. The specific implementation method is as follows:

[0040] First, the laying path of the distributed fiber optic sensing lines is determined based on the spatial structure of the target monitoring area. The target monitoring area is a closed or semi-closed space with potential fire risk, including cable trays, cable bridge passages, integrated utility tunnel compartments, or the interior of energy storage equipment compartments. The laying path is arranged along the potential distribution direction of the main heat sources in the target monitoring area, so that the distributed fiber optic sensing lines spatially cover all possible fire ignition locations. During laying, the distributed optical fibers are fixed to cable supports, equipment housings, or structural surfaces, and a stable contact is maintained between the optical fibers and the monitored structure to ensure stable temperature transfer.

[0041] After the fiber optic cable is laid, one end is connected to a distributed fiber optic temperature measurement device. The device uses optical time-domain reflectometry (OTDR) based on Raman scattering to calculate the temperature. The device emits a pulsed laser signal into the fiber and collects the Stokes-scattered and anti-Stokes-scattered light signals returning along the fiber. The temperature at the corresponding location is calculated by determining the ratio between the intensity of the anti-Stokes-scattered light signal and the intensity of the Stokes-scattered light signal, combined with a calibration coefficient.

[0042] A spatial coordinate system is established along the length of the optical fiber, defining the fiber's starting end as the spatial origin. A continuous sequence of spatial coordinates is then established along the fiber's length. The optical fiber is divided into multiple discrete spatial sampling units, each corresponding to a temperature measurement location. The length of each spatial sampling unit is determined based on the spatial resolution of the temperature measuring device, preferably between 0.5 meters and 1 meter.

[0043] Under stable environmental conditions with no fires or abnormal heat sources in the target monitoring area, temperature data from each spatial sampling unit along the optical fiber was continuously collected multiple times. The collected temperature data were then averaged to eliminate the influence of instantaneous temperature fluctuations on the reference temperature. The number of samplings was set to M, where M ranges from 20 to 50. For the i-th spatial sampling unit along the optical fiber, its initial temperature value was calculated using the following formula: the initial temperature value is the arithmetic mean of the M temperature measurements. After averaging the temperatures of all spatial sampling units, the initial temperature values ​​of each spatial sampling unit were arranged according to their spatial location along the optical fiber, constructing an initial temperature distribution sequence T0(x), where x represents the spatial location along the optical fiber. This initial temperature distribution sequence represents the reference temperature distribution corresponding to each spatial location along the distributed optical fiber under normal environmental conditions. This sequence serves as the basis for subsequent temperature change analysis and anomaly identification.

[0044] In this initial temperature distribution sequence, the independent variable is spatial location, increasing along the fiber direction from the spatial starting point; the dependent variable is the initial temperature value at the corresponding spatial location. The initial temperature distribution sequence obtained in this way can accurately reflect the spatial temperature distribution characteristics of the target monitoring area under stable conditions, providing benchmark reference data for subsequent fire source identification and location tracking calculations.

[0045] During the monitoring process, temperature data along the optical fiber is periodically collected to obtain the temperature distribution sequence Tt(x) at different times, and the temperature increment distribution ΔTt(x) at adjacent times is calculated.

[0046] In this embodiment, temperature data along the optical fiber is periodically collected during the monitoring process to obtain a temperature distribution sequence at different times, and the temperature increment distribution between adjacent time points is calculated. The specific implementation method is as follows:

[0047] First, using the obtained initial temperature distribution sequence as baseline data, the time interval for each temperature data acquisition is set to Δt. The time interval Δt is typically set between 1 and 10 seconds, selected based on the response speed and monitoring accuracy requirements of the fiber optic temperature measurement device. Within each time interval, the fiber optic temperature measurement device sends a pulsed laser signal to the optical fiber, collects the backscattered light signal, and calculates the temperature value of each spatial sampling unit based on the scattered light signal. The acquired temperature data includes the temperature values ​​of all spatial sampling units along the fiber optic line at that moment, and these data constitute a temperature distribution sequence at a given time point.

[0048] Each temperature data acquisition corresponds to a time point t, denoted as Tt(x), where t is the acquisition time and x is the spatial location along the optical fiber. Each Tt(x) represents the temperature value at a specific spatial location on the optical fiber at time point t. x changes continuously from the starting point to the ending point of the optical fiber, covering the entire monitoring area. After each temperature acquisition, the data is transmitted to the data processing unit in real time via the transmission interface.

[0049] After continuously collecting data at multiple time points, the incremental change in temperature data between adjacent time points is calculated to obtain the temperature increment distribution. Specifically, the temperature distribution Tt(x) at a certain time t is calculated compared with the temperature distribution at the previous time t-1. The difference between them yields the temperature increment distribution ΔTt(x) at adjacent time points: Where ΔTt(x) represents the temperature change at position x along the fiber between time t and time t-1. For each spatial sampling unit x on the fiber, the temperature increment at that position is calculated to obtain the temperature increment distribution sequence along the entire fiber.

[0050] In practice, calculating the temperature increment distribution ΔTt(x) requires real-time processing and storage of the collected data at each time point t. To ensure the accuracy and timeliness of the temperature data, a sliding window approach is typically used for temperature increment calculation. For example, after the system collects the temperature data Tt(x) at time t, it directly compares it with the temperature data at time t-1 to calculate the temperature increment ΔTt(x). The temperature increment distribution ΔTt(x) reflects the temperature change trend between different time points, facilitating the identification of abnormal temperature fluctuations or hotspot areas.

[0051] If the temperature increment ΔTt(x) exceeds the set threshold, it is considered that a local heat source change may have occurred in the region, requiring further temperature change analysis and anomaly source tracing. The threshold ΔTthreshold needs to be adjusted based on experimental data to accommodate temperature fluctuations under different environments; a common setting is... to .

[0052] It should be noted that by implementing this step, the continuously acquired and calculated temperature increment distribution ΔTt(x) provides the temperature change characteristics at different locations along the optical fiber at each time point, providing effective data support for subsequent anomaly detection and fire source location identification.

[0053] Based on ΔTt(x), temperature anomaly segments are identified, and the temperature rise gradient characteristics and spatial diffusion width characteristics of the anomaly segments are extracted. Asymmetric temperature rise gradient coefficient Kg and thermal diffusion offset index Kd are generated to form an anomaly feature set B=(Kg,Kd).

[0054] First, based on the obtained temperature increment distribution, the temperature increment values ​​corresponding to each spatial sampling unit along the optical fiber are arranged in spatial order to construct a continuous temperature increment curve. Specifically, the construction method is as follows: using the spatial location along the optical fiber as the abscissa and the temperature increment value corresponding to that location as the ordinate, the temperature increment values ​​of all spatial sampling units are connected in spatial order to form a continuous temperature increment curve reflecting the temperature change trend. This continuous temperature increment curve is used to describe the instantaneous temperature change distribution characteristics along the optical fiber direction in the monitoring area.

[0055] To reduce the impact of environmental noise and random measurement fluctuations on the continuous temperature increment curve, a sliding window smoothing process is applied. The specific steps are as follows: First, the sliding window length is set to three consecutive spatial sampling units, and the window is moved gradually along the fiber optic spatial direction with a step size of one spatial sampling unit. At each sliding window position, the temperature increment values ​​corresponding to the three consecutive spatial sampling units within the window are extracted, and the arithmetic mean of these three temperature increment values ​​is calculated. Then, this arithmetic mean is assigned to the spatial sampling unit position corresponding to the center of the window, thus obtaining the smoothed temperature increment sequence. This sliding window averaging method eliminates single-point anomalous noise, making the temperature change curve more stable.

[0056] After obtaining the smoothed temperature increment sequence, gradient calculations are performed on the temperature change trends between adjacent spatial sampling units. The temperature increment gradient is calculated as follows: the temperature increment difference between two adjacent spatial sampling units is calculated, and this difference is divided by the spatial distance between the two spatial sampling units to obtain the temperature increment gradient value. The distance between the spatial sampling units is determined during the laying of the distributed optical fiber and is a fixed value. Through the above calculations, the temperature increment gradient distribution at various spatial locations along the optical fiber can be obtained.

[0057] To pinpoint potential locations of abnormal temperature rises, a temperature increment gradient threshold needs to be predetermined. This threshold is obtained through statistical calculation of temperature data collected during historical stable operation. The specific steps are as follows: When the target monitoring area is in a normal and stable state, at least thirty sets of temperature increment distribution data are continuously collected, and the temperature increment gradient value in each set is calculated. Then, the mean and standard deviation of all temperature increment gradient values ​​are calculated, and the sum of the mean and three times the standard deviation is used as the temperature increment gradient threshold. When the temperature increment gradient at a certain spatial location exceeds the temperature increment gradient threshold, that spatial location is marked as a candidate anomaly location.

[0058] After identifying candidate anomaly locations, a progressively expanding search is performed outwards from each candidate anomaly location as the center, along the fiber optic spatial direction. During the expanding search, the temperature increment value of adjacent spatial sampling units is checked one by one and compared with a pre-set temperature increment benchmark threshold. The temperature increment benchmark threshold is obtained by statistically analyzing historical stable operating temperature increment data. Specifically, the average of the absolute values ​​of all historical temperature increments is calculated, and then two times the standard deviation is added to this average value to obtain the temperature increment benchmark threshold.

[0059] When the temperature increment values ​​of multiple consecutive spatial sampling units are greater than the temperature increment reference threshold during the extended search process, the spatial intervals corresponding to these consecutive spatial sampling units are determined as temperature anomaly segments. The starting position of the temperature anomaly segment is the location where the temperature increment first exceeds the temperature increment reference threshold, and the ending position is the location where the temperature increment falls below the temperature increment reference threshold again. Through the above steps, continuous temperature anomaly segments with significant temperature rise characteristics can be accurately identified in the temperature increment distribution, providing a reliable data foundation for subsequent ignition point location identification.

[0060] After identifying the temperature anomaly zones, it is necessary to further extract temperature rise gradient features and spatial diffusion width features, and calculate the asymmetric temperature rise gradient coefficient and thermal diffusion shift index based on these features to construct an anomaly feature set. The anomaly feature set consists of the asymmetric temperature rise gradient coefficient and the thermal diffusion shift index, used to characterize the directional asymmetry of temperature changes and the degree of spatial thermal diffusion shift within the temperature anomaly zones.

[0061] First, the asymmetric temperature gradient coefficient Kg is calculated. The specific process is as follows:

[0062] First, the temperature increment sequence within the temperature anomaly zone is processed using empirical mode decomposition (EMD) according to its spatial location. EMD is an adaptive signal decomposition method, and its implementation steps are as follows: First, all local maxima and local minima are found in the temperature increment sequence; then, cubic spline interpolation is used to connect all local maxima to form an upper envelope curve, and all local minima are connected to form a lower envelope curve; subsequently, the average curves of the upper and lower envelope curves are calculated, and the original temperature increment sequence is subtracted from the average curve to obtain a new sequence; the above process is repeated until the obtained sequence satisfies the intrinsic mode function (EMF) condition, that is, the difference between the number of zero crossovers and the number of extreme points in the sequence does not exceed 1, and the local average of the upper and lower envelope curves is zero. Through the above screening process, multiple intrinsic mode components are obtained, and their frequency characteristics are determined according to the average wavelength of each intrinsic mode component. The intrinsic mode component with the largest average wavelength is identified as the low-frequency intrinsic mode component, and this low-frequency intrinsic mode component is used as the temperature rise trend sequence.

[0063] After obtaining the temperature rise trend sequence, the spatial location with the highest temperature value in the sequence is taken as the temperature peak location, and this peak location is used as the center to divide the temperature rise trend sequence into a left temperature rise sequence and a right temperature rise sequence. Then, Hilbert transform is performed on the left and right temperature rise sequences respectively. The Hilbert transform is achieved by convolving the original sequences, and its mathematical expression is: H(x) represents the Hilbert transform result of the temperature rise trend sequence T(x) at spatial location x, and T(τ) represents the function value of the temperature rise trend sequence at spatial location τ, that is, the temperature change value corresponding to the low-frequency intrinsic mode components obtained after empirical mode decomposition. τ represents the integration variable, used to traverse all locations of the temperature rise trend sequence throughout the entire spatial range. This is the normalization constant for the Hilbert transform, used to ensure that the amplitude of the transformed signal maintains the correct proportional relationship. This yields the corresponding analytic signal. Where j represents the imaginary unit, and T(x) represents the function value of the temperature rise trend sequence at spatial location x, i.e., the temperature change value corresponding to the low-frequency intrinsic mode components obtained through empirical mode decomposition. Furthermore, the instantaneous amplitude of the analytical signal is calculated. The instantaneous amplitude curves of the left and right temperature rise sequences were obtained using this method.

[0064] After obtaining the instantaneous amplitude curve, the gradient of the instantaneous amplitude curve is calculated along the spatial direction. Specifically, the instantaneous amplitude difference between adjacent spatial sampling units is calculated, and this difference is divided by the spatial distance between the sampling units to obtain the spatial gradient value at the corresponding location. Then, the spatial gradient values ​​of the left and right temperature rise sequences are averaged to obtain the left temperature rise gradient. With the temperature gradient on the right .

[0065] Obtain the temperature gradient on the left side With the temperature gradient on the right Then, the asymmetric temperature rise gradient coefficient is calculated by the ratio of the difference between the two to the sum. The asymmetric temperature rise gradient coefficient is calculated as follows: This coefficient is used to characterize the degree of spatial asymmetry in temperature rise within an abnormal temperature zone.

[0066] The thermal diffusion displacement index Kd is calculated as follows:

[0067] First, the temperature increment values ​​corresponding to each spatial sampling unit within the temperature anomaly zone are extracted and arranged in spatial order. To construct a probability distribution, the temperature increment values ​​need to be normalized. Specifically, the temperature increment value of each spatial sampling unit is divided by the sum of all temperature increment values ​​within the entire anomaly zone, thus obtaining the corresponding probability density value. This method allows the construction of an actual temperature diffusion distribution function, which represents the spatial distribution of temperature energy.

[0068] After obtaining the actual temperature diffusion distribution function, the spatial location with the largest temperature increment within the temperature anomaly zone is taken as the temperature peak location. Using this temperature peak location as the center of symmetry, a reference temperature diffusion distribution function is constructed based on the laws of heat conduction and diffusion. The reference temperature diffusion distribution function adopts a Gaussian distribution form symmetrical about the temperature peak location, and its probability density value is calculated using a Gaussian function, expressed as: In the formula, Represents the spatial location of the reference temperature diffusion distribution function. The probability density value at that location; σ represents the location of the temperature peak; σ represents the standard deviation of the Gaussian distribution, where the standard deviation of the Gaussian function is determined by half of the spatial diffusion width of the temperature anomaly segment, thus ensuring that the reference temperature diffusion distribution function and the actual temperature diffusion distribution function have the same spatial range.

[0069] Then, the Wasserstein distance between the actual temperature diffusion distribution function and the reference temperature diffusion distribution function is calculated. The Wasserstein distance represents the minimum transmission cost required to transform one probability distribution into another. The specific calculation steps are as follows: First, the cumulative distribution functions of the two probability distributions at spatial locations are calculated, expressed as: In the formula, This represents the cumulative distribution function value of the actual temperature diffusion distribution function, i.e., from the starting position of the temperature anomaly segment to the spatial position. The summation of all probability density values ​​is used to describe the cumulative distribution of actual temperature energy in the spatial direction. This represents the actual temperature diffusion distribution function in space. The probability density value at that point is calculated by normalizing the temperature increment. This represents the cumulative distribution function value of the reference temperature diffusion distribution function, i.e., the reference Gaussian distribution in space. The previous probability density cumulative quantity was used to describe the cumulative spatial distribution of temperature energy under ideal symmetric thermal diffusion conditions. Represents the spatial location of the reference temperature diffusion distribution function. The probability density value at each location is calculated using a Gaussian distribution function centered on the temperature peak location. Then, the difference between the two cumulative distribution functions at each spatial location is calculated; finally, the absolute values ​​of the differences at all spatial locations are integrated and summed to obtain the minimum transmission distance, i.e., the Wasserstein distance. After obtaining the Wasserstein distance, it is normalized to the spatial diffusion width of the temperature anomaly segment. The spatial diffusion width is defined as the spatial distance between the starting and ending locations of the temperature anomaly segment. By dividing the Wasserstein distance by the spatial diffusion width, the thermal diffusion offset index is obtained. This index characterizes the degree of spatial offset of the thermal diffusion distribution within the temperature anomaly segment relative to the ideal symmetrical diffusion state.

[0070] Finally, the calculated asymmetric temperature rise gradient coefficient and thermal diffusion offset index are combined to form an abnormal feature set B=(Kg,Kd), which is used for subsequent ignition point location identification and dynamic tracking analysis.

[0071] A thermal diffusion evolution model is constructed based on the abnormal feature set B, and the reverse heat source fitting calculation is performed on the abnormal section to obtain the set of candidate ignition point locations.

[0072] After obtaining the set of anomalous features, it is necessary to construct a thermal diffusion evolution model using the asymmetric temperature rise gradient coefficient and thermal diffusion offset index in the set of anomalous features, and determine the set of candidate ignition point locations by fitting the reverse heat source.

[0073] First, a spatial temperature gradient field is established based on the numerical relationship between the asymmetric temperature rise gradient coefficient and the thermal diffusion shift index in the anomaly feature set. Specifically, within the temperature anomaly zone, a discrete temperature field is constructed using the spatial location of each spatial sampling unit as the abscissa and the temperature increment value of the corresponding spatial sampling unit as the ordinate. Then, the spatial temperature gradient between each spatial sampling unit is calculated based on the discrete temperature field. The spatial temperature gradient is obtained by calculating the ratio of the temperature difference between adjacent spatial sampling units to their spatial distance. The asymmetric temperature rise gradient coefficient is used as a direction correction factor to weight and adjust the left and right spatial gradients, thus forming a spatial temperature gradient field with direction shift characteristics. Based on this, a thermal diffusion evolution equation is constructed. The thermal diffusion evolution equation adopts the form of a one-dimensional partial differential equation of heat conduction, which is expressed as the rate of temperature change with time equal to the product of the thermal diffusion coefficient and the second derivative of the spatial temperature. To facilitate calculation, the equation is discretized according to spatial sampling units, and the spatial second-order difference term is calculated using the finite difference method, thereby obtaining the thermal diffusion evolution equation.

[0074] After obtaining the heat diffusion evolution equation, the asymmetric temperature rise gradient coefficient and the thermal diffusion shift index are introduced into the thermal diffusion evolution equation as correction parameters. Specifically, the asymmetric temperature rise gradient coefficient is applied to the spatial gradient term to weight the direction of spatial temperature change, resulting in different diffusion rates in the asymmetric direction. Simultaneously, the thermal diffusion shift index is added to the diffusion equation as a spatial drift constraint parameter, shifting the diffusion center towards the thermal diffusion shift direction by adding a shift correction term to the spatial coordinates. After introducing the parameters, the temperature increment distribution at the current moment in the temperature anomaly segment is used as the initial condition. Iterative calculations are performed at each time step to obtain the predicted temperature diffusion distribution under different time conditions. The time step is determined based on the thermal diffusion stability condition, and its value is the square of the distance between adjacent spatial sampling units divided by twice the thermal diffusion coefficient.

[0075] Error fitting calculations were performed between the measured temperature diffusion distribution and the predicted temperature diffusion distribution calculated by the model. The error calculation method was as follows: at each spatial sampling unit location, the difference between the actual temperature increment and the predicted temperature increment was calculated, and the squares of the differences for all spatial sampling units were summed to obtain the sum of squared errors. To determine the most likely heat source location, different spatial sampling units within the temperature anomaly zone were successively assumed as potential heat source locations, and the thermal diffusion evolution calculation and error sum of squares calculation were repeatedly performed for each assumed heat source location. The spatial location with the smallest error sum of squares was selected as the optimal fitted heat source location using the least squares inversion method. Simultaneously, an error threshold was set, which was obtained through statistical analysis of fitting errors using historical normal temperature data; specifically, it was the average historical fitting error plus twice the standard deviation.

[0076] When the sum of squared errors corresponding to a hypothetical heat source location is less than the error threshold, that spatial location is identified as a candidate heat source location that meets the condition. After performing the above reverse search on all spatial sampling units within the temperature anomaly zone, multiple candidate heat source locations that meet the error threshold condition can be obtained.

[0077] Finally, all candidate heat source locations that meet the error threshold are organized and combined according to their spatial location to form a set of candidate ignition point locations. This set of candidate ignition point locations is used for subsequent ignition point trajectory correlation analysis and dynamic tracking calculations.

[0078] Trajectory correlation analysis is performed on the set of candidate ignition point locations obtained at multiple consecutive time points to select the point with the smallest spatial drift and continuous temperature increase as the target ignition point, and the precise location and dynamic trajectory of the target ignition point are output.

[0079] After obtaining a set of candidate ignition point locations at multiple consecutive time points, it is necessary to determine the actual ignition location through trajectory correlation analysis and output the dynamic trajectory of the ignition point changing over time.

[0080] Spatial location matching is performed on the set of candidate ignition point locations obtained from multiple consecutive time points. Specifically, the candidate ignition point locations at each monitoring time are recorded according to the spatial coordinates along the optical fiber, and a time-series location set is constructed in chronological order. For candidate ignition point locations at two adjacent monitoring times, the spatial distance between the two locations is calculated. The spatial distance is calculated as the absolute difference between the two spatial coordinates. Then, all calculated spatial distances are combined according to their corresponding candidate locations to form a spatial distance matrix. To establish a temporal correlation, a correlation distance threshold needs to be set. The correlation distance threshold is determined based on the length of the distributed optical fiber spatial sampling unit, typically set to the length of two spatial sampling units. When the spatial distance is less than the correlation distance threshold, the candidate ignition point locations at two time points are considered to have spatial continuity, and a temporal correlation is established between the corresponding candidate locations.

[0081] After establishing temporal correlations, the candidate ignition point locations at different monitoring times are spliced ​​together based on these correlations. Specifically, the earliest monitoring time's candidate ignition point location is used as the starting position, and candidate locations with temporal correlations are connected sequentially in chronological order to form multiple candidate ignition trajectories. For each candidate ignition trajectory, the spatial position change between adjacent monitoring times is further calculated. The spatial position change is obtained by calculating the spatial distance between corresponding locations at two adjacent times. Then, all spatial position changes are arranged in chronological order to form a spatial drift sequence, and the average value of this sequence is calculated as the average spatial drift of the candidate ignition trajectory.

[0082] While calculating the spatial drift sequence, it is necessary to extract temperature change information at the corresponding spatial locations of the candidate fire trajectories. The specific method is as follows: based on the spatial location corresponding to each moment in the candidate fire trajectory, the temperature increment value at that spatial location is extracted from the temperature distribution sequence, and a temperature increment change sequence is formed in chronological order. Then, a continuous growth rate is calculated for the temperature increment change sequence. The continuous growth rate of temperature increment is obtained by calculating the ratio of the temperature increment difference between adjacent moments to the temperature increment of the previous moment. When the temperature increment difference at multiple consecutive moments is greater than zero, the candidate fire trajectory is considered to have a continuous temperature rise characteristic.

[0083] A comprehensive comparative analysis was conducted on all candidate fire trajectories. Specifically, the average spatial drift and corresponding continuous temperature increment growth rate were calculated for each candidate fire trajectory, and they were sorted in ascending order of average spatial drift. Among the candidate fire trajectories that met the condition of continuous temperature increment growth, the trajectory with the smallest average spatial drift was selected as the target fire trajectory. The spatial position of this target fire trajectory at each monitoring time is the spatial position sequence of the target fire point. By recording the change of this spatial position sequence over time, the dynamic trajectory of the target fire point can be obtained, and the precise spatial position of the target fire point can be output.

[0084] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for precise tracking of fire ignition points based on distributed optical fiber, characterized in that: include: Distributed optical fiber sensing lines are laid along the target monitoring area to obtain the corresponding initial temperature distribution sequence T0(x), where x is the spatial position along the optical fiber line; During the monitoring process, temperature data along the optical fiber is periodically collected to obtain the temperature distribution sequence Tt(x) at different times, and the temperature increment distribution ΔTt(x) between adjacent times is calculated. Based on ΔTt(x), temperature anomaly segments are identified, and the temperature rise gradient characteristics and spatial diffusion width characteristics of the anomaly segments are extracted. Asymmetric temperature rise gradient coefficient Kg and thermal diffusion offset index Kd are generated to form an anomaly feature set B=(Kg,Kd). The calculation of the asymmetric temperature rise gradient coefficient includes the following steps: Empirical mode decomposition is performed on the temperature increment sequence within the temperature anomaly zone according to spatial location. Multiple intrinsic mode components are separated by successively screening local extreme points and constructing envelope curves. The low-frequency intrinsic mode component containing the main temperature rise trend is selected as the temperature rise trend sequence. Using the temperature peak position of the temperature rise trend sequence as the center, the temperature rise trend sequence is divided into a left temperature rise sequence and a right temperature rise sequence. Hilbert transforms are performed on the left and right temperature rise sequences respectively to obtain the corresponding instantaneous amplitude curves. The spatial gradient values ​​of the left and right temperature rise sequences are calculated based on the instantaneous amplitude curves, and the left and right temperature rise gradients are obtained respectively. The asymmetric temperature rise gradient coefficient is constructed based on the ratio of the difference between the left and right temperature rise gradients to their sum. The calculation of the thermal diffusion offset index includes the following steps: extracting the temperature increment values ​​corresponding to each spatial sampling unit within the temperature anomaly segment, and normalizing them according to their spatial location to convert the temperature increment values ​​into a probability density distribution, thereby constructing an actual temperature diffusion distribution function; using the temperature peak position of the temperature anomaly segment as the center of symmetry, constructing a reference temperature diffusion distribution function symmetrical about the temperature peak position according to the heat conduction and diffusion law; calculating the optimal transmission cost between the actual temperature diffusion distribution function and the reference temperature diffusion distribution function based on the spatial location difference between them, and obtaining the Wasserstein distance value by solving for the minimum transmission distance; normalizing the Wasserstein distance value and the spatial diffusion width of the temperature anomaly segment to obtain the thermal diffusion offset index characterizing the degree of spatial offset of thermal diffusion. A thermal diffusion evolution model is constructed based on the abnormal feature set B, and the reverse heat source fitting calculation is performed on the abnormal section to obtain the set of candidate ignition point locations. Trajectory correlation analysis is performed on the set of candidate ignition point locations obtained at multiple consecutive time points to select the point with the smallest spatial drift and continuous temperature increase as the target ignition point, and the precise location and dynamic trajectory of the target ignition point are output.

2. The method for precise tracking of fire ignition points based on distributed optical fiber according to claim 1, characterized in that: A spatial coordinate system is established along the length of the optical fiber. The starting end of the optical fiber is defined as the spatial starting point. A continuous spatial coordinate sequence is established along the length of the optical fiber, and the optical fiber is divided into multiple discrete spatial sampling units. Each spatial sampling unit corresponds to a temperature measurement position. For the i-th spatial sampling unit along the optical fiber, the initial temperature value is calculated. The initial temperature values ​​of each spatial sampling unit are arranged according to the spatial position of the optical fiber to construct the initial temperature distribution sequence T0(x), where x is the spatial position along the optical fiber.

3. The method for precise tracking of fire ignition points based on distributed optical fiber according to claim 1, characterized in that: The temperature increment distribution ΔTt(x) between adjacent time points represents the temperature change at position x along the fiber between time t and time t-1.

4. The method for precise tracking of fire ignition points based on distributed optical fiber according to claim 1, characterized in that: Identifying temperature anomaly zones based on temperature increment distribution includes the following steps: A continuous temperature increment curve is constructed based on the temperature change values ​​at each spatial location in the temperature increment distribution. The continuous temperature increment curve is then smoothed using a sliding window. A smooth temperature increment sequence is formed by taking the arithmetic mean of the temperature increments within the window. In the smoothed temperature increment sequence, the temperature increment gradient between adjacent spatial sampling units is calculated, and the positions where the temperature increment gradient is greater than a preset gradient threshold are marked as candidate anomaly positions. The search is gradually expanded to both sides of the candidate anomaly location. When the temperature increment of the continuous spatial sampling units is greater than the temperature increment reference threshold, the continuous interval is determined as the temperature anomaly segment.

5. The method for precise tracking of fire ignition points based on distributed optical fiber according to claim 1, characterized in that: The thermal diffusion evolution model is constructed based on the anomalous feature set B, including the following steps: Based on the numerical relationship between the asymmetric temperature rise gradient coefficient and the thermal diffusion migration index in the anomalous feature set, a temperature spatial gradient field is established, and the thermal diffusion evolution equation is constructed using the temperature increment value of each spatial sampling unit in the temperature anomaly section as the initial condition. The thermal diffusion evolution equation is obtained by discretization calculation of the one-dimensional thermal conduction partial differential equation.

6. The method for precise tracking of fire ignition points based on distributed optical fiber according to claim 5, characterized in that: An asymmetric temperature rise gradient coefficient is introduced as a gradient correction factor in the thermal diffusion evolution equation, and a thermal diffusion offset index is introduced as a spatial drift constraint parameter. The temperature field is iteratively evolved to obtain the predicted temperature diffusion distribution at multiple time points. By minimizing the error between the actual temperature diffusion distribution and the predicted temperature diffusion distribution, the least squares inversion method is used to progressively search for possible heat source locations, resulting in several candidate heat source locations that meet the error threshold condition. The candidate heat source locations are then aggregated according to their spatial location to form a set of candidate ignition point locations.

7. The method for precise tracking of fire ignition points based on distributed optical fiber according to claim 1, characterized in that: Trajectory correlation analysis is performed on the set of candidate ignition point locations obtained at multiple consecutive time points, including the following steps: Spatial location matching is performed on the set of candidate ignition point locations obtained from multiple consecutive time moments in chronological order. A candidate location association matrix is ​​constructed by calculating the spatial distance between candidate ignition points at adjacent time moments, and a temporal association relationship is established for candidate locations whose spatial distance is less than a preset association distance threshold. Based on the temporal association relationship, the trajectories of candidate ignition points at consecutive time moments are spliced ​​to generate multiple candidate ignition trajectories. The spatial location change amount at adjacent time moments is calculated for each candidate ignition trajectory to obtain the corresponding spatial drift amount sequence. While calculating the spatial drift amount sequence, the temperature increment change sequence at the corresponding spatial location is extracted, and the continuous growth rate of the temperature increment change sequence is used to determine the continuous temperature rise characteristic.

8. The method for precise tracking of fire ignition points based on distributed optical fiber according to claim 7, characterized in that: By comprehensively comparing the average spatial drift and continuous temperature rise rate of each candidate ignition trajectory, the candidate ignition trajectory with the smallest average spatial drift and continuous temperature rise is selected, and the spatial location corresponding to the candidate ignition trajectory is determined as the target ignition point location. At the same time, the dynamic change trajectory of the target ignition point over time is output.

Images