A fabric flame retardant analysis method based on infrared thermal imaging sensing data

Abstract

The application relates to the field of textile detection and discloses a fabric flame resistance analysis method based on infrared thermal imaging sensing data, which comprises the following steps: non-contact temperature measurement on a fabric sample in a controlled thermal excitation field, acquisition of an infrared imaging signal sequence, energy projection rearrangement on the infrared imaging signal sequence; based on the discontinuous change of the energy flow density in a primary thermal behavior field, a boundary disturbance model of a local overheating area is established, the morphological difference between a thermal expansion path and a radiation energy decay area is compared, the thermal potential difference of each area in a critical response graph is mapped into the energy constraint state inside the material, the time bias distribution of energy conduction is deduced, a spectral equalization function is introduced under the time bias distribution, asymmetric compensation is performed on the local abnormal points of the infrared response data, based on a decomposition threshold dynamic track, the flame resistance response equalization rate and the energy flow retention rate are calculated, and a flame resistance performance evaluation index is formed. The application has the advantage of improving the accuracy of flame resistance detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of textile detection, in particular to a fabric flame retardant analysis method based on infrared thermal imaging sensing data. BACKGROUND

[0002] With the wide application of functional textile materials in public transportation, aerospace, protective clothing and other scenes, the flame retardant performance detection of fabrics has become an important part of material safety evaluation. The existing flame retardant test methods are mostly based on combustion experiments under constant heat source or contact type temperature sensor measurement, such as oxygen index method, vertical burning method, etc. Such methods need to perform destructive tests on samples, and it is difficult to obtain the continuous distribution information of heat diffusion, local carbonization and thermal response evolution of the fabric during the combustion process. Especially in light, thin and multi-layer fabrics or composite coating materials, the temperature gradient changes have strong spatial heterogeneity, and the traditional point measurement or time discrete sampling means cannot accurately reflect the thermal conduction and thermal instability characteristics of the material in the early stage of heating, resulting in lag and error in the evaluation results of flame retardant performance. How to obtain the thermal distribution dynamic data of the fabric combustion process under the conditions of non-contact, real-time and spatial continuity, and based on these infrared thermal imaging sensing information, to carry out multi-dimensional physical characteristic analysis and quantitative evaluation on the flame retardant performance, has become a technical problem to be solved in the field of flame retardant detection. Therefore, it is necessary to design a fabric flame retardant analysis method based on infrared thermal imaging sensing data to improve the accuracy of flame retardant detection. SUMMARY

[0003] In view of the shortcomings of the prior art, the present application provides a fabric flame retardant analysis method based on infrared thermal imaging sensing data, which has the advantage of improving the accuracy of flame retardant detection and solving the problems in the above background art.

[0004] To achieve the above purpose of improving the accuracy of flame retardant detection, the present application provides the following technical scheme: a fabric flame retardant analysis method based on infrared thermal imaging sensing data, comprising the following steps:

[0005] Non-contact temperature measurement is performed on the fabric sample in a controlled thermal disturbance field, and an infrared imaging signal sequence is obtained. The infrared imaging signal sequence is energy projected and rearranged, the radiation energy of each pixel is converted into a heat flow migration vector, and a primary thermal behavior field representing the local heat conduction response of the fabric is formed;

[0006] Based on the non-continuous change of energy flow density in the primary thermal behavior field, a boundary disturbance model of the local overheating domain is established. By comparing the shape difference between the heat propagation path and the radiation energy decay area, the heat transition mode of the combustion initiation zone is identified, and a critical response graph reflecting the thermal instability trend is generated;

[0007] The thermal potential difference of each region in the critical response map is mapped to the energy constraint state inside the material, and the time bias distribution of energy conduction is derived by combining the fabric organization density and thermal resistance gradient parameters;

[0008] An asymmetric compensation is performed on the local abnormal points of the infrared response data by introducing a spectral equalization function under the time bias distribution, a fabric combustion stability indicator factor is constructed by using multi-dimensional disturbance residuals, and a dynamic trajectory of the demarcation threshold between the stable region and the unstable region is obtained;

[0009] Based on the decomposition threshold dynamic trajectory, the flame retardant response equalization rate and the energy flow retention rate are calculated to form a flame retardant performance evaluation index, and the flame retardant performance of the fabric is quantitatively graded and judged by comparing the index fluctuation range under different thermal excitation levels.

[0010] Preferably, the process of forming the primary thermal behavior field representing the local thermal conduction response of the fabric is as follows:

[0011] A multi-channel infrared acquisition unit is arranged in the controlled thermal excitation disturbance field to perform time sequence scanning on the surface of the fabric sample;

[0012] Energy integration and temperature difference layering processing are performed on the continuously acquired infrared pixel data, and the energy mutation fragments appearing between adjacent frames are defined as transient thermal conduction events;

[0013] The local heat flow direction is calculated by the energy gradient vector between pixels, and the heat flow vector is spatially rearranged in combination with the fabric texture distribution;

[0014] The local thermal conduction response function domain is constructed by taking the heat flow vector density and the gradient change rate as the core parameters, and the primary thermal behavior field representing the local thermal conduction response of the fabric is formed.

[0015] Preferably, the process of establishing the boundary disturbance model of the local overheating domain is as follows:

[0016] The modulus sequence of the heat flow vector in the primary thermal behavior field is continuously detected to mark the gradient mutation points and their neighborhood distribution;

[0017] A local energy flow disturbance function is constructed in the neighborhood range of the mutation point, and the energy flow abnormal aggregation region is defined as a potential overheating domain;

[0018] The geometric boundary of the overheating domain is determined by using a region growing algorithm based on topological adjacency, and the boundary disturbance coefficient is calculated;

[0019] A boundary disturbance model representing the degree of local thermal imbalance is formed by coupling analysis of the disturbance coefficient and the energy expansion rate.

[0020] Preferably, the process of generating the critical response map reflecting the thermal instability trend is as follows:

[0021] According to the boundary perturbation model, an energy flow propagation path is extracted, and a corresponding sequence is formed between the propagation direction of each path and the energy decay rate;

[0022] A morphological difference matching algorithm is used to compare the spatial consistency of the propagation path and the radiation energy decay region, and a local energy flow inversion region is selected;

[0023] In the energy flow inversion region, the mutation point of the heat propagation rate and the energy flow inversion interval are counted, and the heat transition mode of the combustion initiation region is identified;

[0024] According to the distribution density of the heat transition mode, a heat instability trend surface is generated, and is superimposed on the primary heat behavior field to form a critical response map.

[0025] Preferably, the process of mapping the thermal potential difference of each region in the critical response map to the energy constraint state inside the material is:

[0026] The critical response map is divided into several equipotential energy regions, and the thermal potential distribution is fitted based on the infrared gray balance value and the thermal conductivity gradient;

[0027] In each equipotential energy region, a tissue density parameter and a fiber arrangement direction vector are introduced to perform scale adjustment processing on the thermal potential difference;

[0028] An energy constraint function is constructed by the directional gradient of the thermal potential difference and the energy dissipation rate, reflecting the local energy closure or leakage state of the fabric;

[0029] The global distribution of the energy constraint function forms an energy constraint mapping model inside the material.

[0030] Preferably, the process of deriving the time bias distribution of energy conduction is:

[0031] Perform time series layering processing on the energy constraint mapping model to extract the relative conduction time delay of each energy channel;

[0032] The time delay offset rate between energy channels is analyzed through a sliding regression window to construct a time bias association structure;

[0033] The time bias association structure is parameter balanced and corrected in combination with the fabric thickness, thermal conductivity and surface roughness, and the time bias distribution is output.

[0034] Preferably, the process of performing asymmetric compensation on the local abnormal points of the infrared response data is:

[0035] The infrared response data is unfolded into a spectral energy sequence under the constraint of the time bias distribution;

[0036] According to the local variance of the energy sequence, the abnormal point region is determined, and the abnormal point is corrected by using an adaptive asymmetric compensation function;

[0037] The time sequence residual vector is used to generate a multi-dimensional disturbance residual, and a disturbance principal feature is extracted through principal component analysis, and a combustion stability indicator is constructed based on the disturbance principal feature.

[0038] Preferably, the process of obtaining the dynamic trajectory of the demarcation threshold between the steady state region and the unstable region is:

[0039] The time sequence change of the combustion stability indicator is subjected to clustering segmentation, and a characteristic mutation point and a corresponding energy threshold are calibrated;

[0040] Based on the clustering result, the energy distribution difference between the steady state region and the unstable region is calculated, and a dynamic evolution curve reflecting the heat response transition law is generated;

[0041] The dynamic evolution curve is subjected to trend tracking and dynamic weight balance processing by using a sliding window integral algorithm, and a dynamic trajectory of the demarcation threshold between the steady state region and the unstable region is formed.

[0042] Preferably, the process of forming the flame retardant performance evaluation index is:

[0043] The energy flow average value and the fluctuation amplitude of the steady state region and the unstable region are extracted in the demarcation threshold dynamic trajectory;

[0044] The flame retardant response balance rate is calculated by calculating the ratio of the energy flow in the steady state region to the overall energy input;

[0045] The energy flow retention rate is calculated by taking the ratio of the energy decay rate in the unstable region to the energy retention time in the steady state region;

[0046] The flame retardant response balance rate and the energy flow retention rate are fused by weighting to generate a comprehensive evaluation index of the flame retardant performance.

[0047] Preferably, the process of quantitatively grading and determining the flame retardant performance of the fabric is:

[0048] According to the comprehensive evaluation index of the flame retardant performance, a flame retardant grading model containing multiple threshold values is established;

[0049] The fluctuation range of the evaluation index under different heat excitation levels is matched with the model threshold value, and the flame retardant grade of the fabric is obtained;

[0050] The heat response mode and the energy flow retention characteristics of samples of each grade are compared and analyzed, the distinguishing effectiveness of the model is verified, and the quantitative grading and determination results of the flame retardant performance are output.

[0051] Compared with the prior art, the present application provides a fabric flame retardant analysis method based on infrared thermal imaging sensing data, which has the following beneficial effects:

[0052] The application realizes the whole process analysis from the surface thermal radiation distribution to the internal energy conduction state by introducing the multi-dimensional dynamic analysis mechanism of infrared thermal imaging sensing data, can accurately reflect the transient heat conduction characteristics and combustion initiation behavior of the fabric without destroying the sample structure. Through energy projection rearrangement and the construction of heat flow migration vector, the spatial resolution and time sensitivity of thermal response are significantly improved; combined with the boundary disturbance model of local overheating area and thermal transition mode recognition, the early instability signal before fabric combustion can be captured, and the precursor warning of flame retardant failure is realized. Through the time bias distribution derivation of energy conduction and the spectral equalization function compensation, the dynamic noise and local overexposure effect in the infrared signal are effectively suppressed, so that the heat flow information extraction is more stable and reliable. Using the dual index system of combustion stability indicator and flame retardant response equalization rate, the flame retardant performance of different materials under multi-level thermal excitation conditions is adaptively classified and quantitatively evaluated, not only improves the scientificity and repeatability of the evaluation results, but also provides a high-precision, non-contact and expandable analysis method for intelligent flame-retardant fabric research and rapid screening. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 The schematic diagram of the method of the application is shown. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0055] Embodiment 1: please refer to Figure 1 The fabric flame retardant analysis method based on infrared thermal imaging sensing data in the embodiments of the application includes the following steps:

[0056] S1: non-contact temperature measurement of the fabric sample in a controlled thermal excitation field, acquiring an infrared imaging signal sequence, energy projection rearrangement is performed on the infrared imaging signal sequence, the radiation energy of each pixel is converted into a heat flow migration vector, and a primary thermal behavior field representing the local heat conduction response of the fabric is formed.

[0057] The process of forming the primary thermal behavior field representing the local heat conduction response of the fabric in S1 is:

[0058] A multi-channel infrared acquisition unit is arranged in a controlled thermal excitation disturbance field to perform time sequence scanning on the surface of the fabric sample; a controlled thermal excitation disturbance field is built under constant environmental temperature and humidity conditions, and a thermal excitation of a set intensity and duration is applied to the surface of the fabric sample to form a repeatable temperature disturbance field; the infrared acquisition unit includes a plurality of distributed sensing modules, each module is equipped with an infrared detector of different waveband to capture the radiation energy distribution of different areas of the fabric surface, and a multi-channel parallel sampling mode is adopted to obtain a continuous thermal image frame sequence with millisecond time resolution; during the time sequence scanning process, each frame of image records the surface temperature field change of the fabric under the action of instantaneous thermal load, and the dynamic capture of the thermal response process is realized;

[0059] Energy integration and temperature difference layering processing are performed on the continuously acquired infrared pixel data, and an energy mutation segment appearing between adjacent frames is defined as a transient heat conduction event; pixel-by-pixel energy integration operation is performed on the acquired infrared image sequence, that is, the radiation intensity change of each pixel point in the time dimension is accumulated to obtain the heat flux change curve in unit time, and the surface area of the fabric is divided into a plurality of hierarchical intervals according to the temperature gradient through the temperature difference layering algorithm to identify the energy transition layer in the heat conduction process; for a pixel group with significant radiation energy change between adjacent frames, a transient heat conduction event is determined through threshold dynamic matching and time window analysis, which not only reveals the transmission path of thermal disturbance in the fabric, but also describes the local differences of heat diffusion speed and energy release rate;

[0060] The local heat flow direction is calculated through the energy gradient vector between pixels, and the heat flow vector is spatially rearranged in combination with the fabric texture distribution; after identifying the transient heat conduction event, energy gradient operation is performed on the corresponding region to calculate the energy change rate and directionality between adjacent pixels, thereby forming a gradient vector field representing the heat diffusion trend, each vector represents the main direction and intensity of heat energy transmission in the local region, and the vector field is structured and corrected in combination with the microstructure characteristics of the fabric (such as the arrangement angle of warp and weft fibers, porosity and thickness distribution) to make the heat flow direction consistent with the actual fiber heat conduction path; the spatial rearrangement operation is realized through a texture mapping algorithm, which can eliminate the errors caused by the infrared sampling angle and surface roughness, so that the heat flow vector reflects the real heat conduction track on a local scale;

[0061] With heat flux vector density and gradient change rate as the core parameters, the local heat conduction response function domain is constructed, and the primary thermal behavior field representing the local heat conduction response of the fabric is formed; the heat flux vector field after spatial rearrangement is taken as the input, the heat flux density (unit area heat flux) and the gradient change rate (the rate of heat flux change over time) of each local area are extracted, and the local heat conduction response function is established based on the core parameters, which reflects the heat conduction inertia, thermal hysteresis effect and thermal diffusion non-uniformity of the fabric under different energy inputs. The heat conduction response functions of each region on the surface of the sample are continuously interpolated and fitted to form a heat conduction response function domain covering the whole region.

[0062] S2: Based on the discontinuous change of energy flux density in the primary thermal behavior field, the boundary disturbance model of the local overheating domain is established, the thermal transition mode of the combustion initiation zone is identified by comparing the shape difference between the heat propagation path and the radiation energy decay area, and the critical response graph reflecting the thermal instability trend is generated.

[0063] The process of establishing the boundary disturbance model of the local overheating domain in S2 is as follows:

[0064] The modulus sequence of the heat flux vector in the primary thermal behavior field is continuously detected, and the gradient mutation points and their neighborhood distribution are marked; in the primary thermal behavior field, the modulus of the heat flux vector of each spatial unit is extracted in time sequence to form a sequence, and the sequence is smoothed to reduce the influence of measurement noise, for example, through short-time averaging or weighted filtering method, in the smoothed sequence, by analyzing the change amplitude and trend between adjacent time points, the positions where the heat flux modulus appears rapid increase or decrease are identified, these positions are the gradient mutation points, according to the distribution of pixels or units around the mutation points, the neighborhood range of each mutation point is determined, and the heat flux modulus change in the neighborhood is recorded;

[0065] The local energy flow disturbance function is constructed in the neighborhood range of the mutation point, and the abnormal energy flow aggregation area is defined as the potential overheating domain; for each marked gradient mutation point, the spatial units in its neighborhood are analyzed by weighting according to the distance from the mutation point and the heat flux direction, the heat flow change intensity in the neighborhood is summarized to form a local energy flow disturbance function, which is used to reflect the energy concentration or diffusion trend in the region, in the disturbance function, the energy accumulation in the abnormal energy flow aggregation area in the neighborhood is judged, and the region with continuous abnormal energy accumulation is defined as the potential overheating domain, the spatial position, area and heat flux intensity change of the potential overheating domain are recorded;

[0066] The geometric boundary of the overheating area is determined by using a region growing algorithm based on topological adjacency, and a boundary disturbance coefficient is calculated; starting from the high-energy unit of the potential overheating area, the surrounding units are gradually expanded through topological adjacency (such as the connection relationship of adjacent pixels or units), and the units meeting the energy concentration and direction consistency conditions are included in the overheating area, until the region growing ends, after the region growing, the outer edge of the region is smoothed to make the boundary form more continuous and reasonable, and the boundary disturbance coefficient is evaluated according to the intensity of the heat flow change on the boundary and the tortuosity of the boundary form, which can quantify the disturbance degree of the boundary and the characteristics of local energy imbalance;

[0067] Through coupling analysis of the disturbance coefficient and the energy expansion rate, a boundary disturbance model representing the degree of local thermal imbalance is formed; the boundary disturbance coefficient of each overheating area is compared and analyzed with the area or energy expansion speed of the region in a continuous time window, a representation model of the degree of regional thermal imbalance is established, the relationship between the boundary disturbance intensity and the energy expansion speed is analyzed, the trend of rapid accumulation or diffusion of thermal energy is identified, and it is judged whether the local thermal area is likely to be unstable or combustion initiation, and the imbalance characteristics of each overheating area are summarized to form a complete boundary disturbance model.

[0068] The process of generating a critical response graph reflecting the thermal instability trend in S2 is:

[0069] According to the boundary disturbance model, the energy flow expansion path is extracted, and the extension direction of each path and the energy decay rate form a corresponding sequence; after the boundary disturbance model is constructed, the energy distribution and heat flow direction of each overheating area in the model are analyzed, the local energy expansion path is extracted along the heat flow vector, each path includes a sequence of spatial units along the heat flow direction, and the energy change of each unit is recorded, the directional sequence of energy flow expansion is formed by recording the time sequence of the thermal energy decay rate of each unit on the path, which can accurately reflect the propagation trend and decay characteristics of heat in the local material;

[0070] The spatial consistency of the expansion path and the radiation energy decay area is compared by using a morphological difference matching algorithm, and the local energy flow inversion area is screened out; the extracted energy flow expansion path is compared with the observed radiation energy decay area in the infrared imaging data, the coincidence degree, deviation angle and energy difference of the path and the decay area are analyzed by using a morphological difference matching algorithm, and the area where the expansion path and the radiation energy decay area exist obvious deviation or reverse energy flow is marked as a local energy flow inversion area, and the abnormal area of thermal energy expansion is identified by comparing the directionality, energy change trend and spatial position consistency of the path;

[0071] The mutation point of the heat propagation rate in the energy flow inversion region and the energy flow reversal region are identified to recognize the heat transition mode of the combustion initiation region; in the calibrated local energy flow inversion region, the change of the heat flow rate of each unit is analyzed in time sequence to identify the mutation point of the heat propagation rate, which usually corresponds to the position of local energy accumulation or heat rebound, and the length and intensity of the energy flow direction reversal are counted, and these information are combined to form the heat transition mode of the combustion initiation region, which can describe the characteristics of local heat energy sudden accumulation or direction reversal and provide quantitative basis for judging the potential combustion risk;

[0072] A heat instability trend surface is generated according to the distribution density of the heat transition mode, and is superimposed on the primary thermal behavior field to form a critical response map; the heat transition mode of the combustion initiation region is statistically analyzed in terms of spatial distribution density, and a three-dimensional heat instability trend surface is generated according to the mode density and energy intensity, which reflects the distribution trend of local thermal imbalance and potential combustion risk, and the distribution trend surface is superimposed on the primary thermal behavior field to form a complete critical response map by combining with the original heat flow vector and local heat conduction response information, which not only shows the spatial distribution of the heat instability trend, but also intuitively reflects the intensity and possible spread path of the local energy flow anomaly.

[0073] S3: mapping the thermal potential difference of each region in the critical response map to the energy constraint state inside the material, combining the fabric organization density and thermal resistance gradient parameters, and deriving the time bias distribution of energy conduction.

[0074] The process of mapping the thermal potential difference of each region in the critical response map to the energy constraint state inside the material in S3 is as follows:

[0075] The critical response map is divided into several equipotential energy regions, and the thermal potential distribution is fitted based on the infrared gray balance value and the thermal conductivity gradient; the image is subjected to multi-scale spatial segmentation, and the entire fabric surface is divided into several regions, the internal thermal energy difference of each region is small, and an equipotential energy region is preliminarily formed, the infrared gray value in each equipotential energy region is statistically analyzed, the local thermal peak and thermal valley positions are identified through gray histogram analysis, and the thermal conductivity gradient is evaluated by combining the heat flow change trend between consecutive frames, based on these gray and gradient information, a weighted fitting method is used to generate the thermal potential distribution surface of each equipotential energy region, so that the thermal potential distribution can truly reflect the energy accumulation and diffusion state inside the region;

[0076] The thermal potential difference is rescaled by introducing the tissue density parameter and the fiber arrangement direction vector in each equipotential energy region; the microstructure characteristics of the fabric, including the fiber arrangement direction, the fabric density, and the fiber interlacing mode, are introduced into the thermal potential analysis for each equipotential energy region, the spatial scale of the local thermal potential is adjusted by analyzing the conduction efficiency of the heat flow along the main direction and the secondary direction of the fiber, and the damping effect of the tissue density on the heat diffusion, so that the thermal potential value not only reflects the infrared imaging signal, but also reflects the actual influence of the fiber structure on the heat conduction, the local neighborhood window analysis method is used to fuse the thermal potential of each pixel with its adjacent pixels and the fiber direction information, so as to enhance the physical rationality of the thermal potential on the microstructure, and a fine-grained thermal potential distribution map is generated;

[0077] An energy constraint function is constructed by the directional gradient of the thermal potential difference and the energy dissipation rate to reflect the local energy closure or leakage state of the fabric; after the local thermal potential adjustment is completed, the directional gradient of the thermal potential difference in each equipotential energy region is analyzed to determine the heat flow trend along the fiber direction and across the fiber direction, the local energy dissipation rate is calculated by counting the heat energy attenuation speed in the continuous time window, and the energy constraint function is constructed by combining the thermal potential gradient and the energy dissipation rate to depict the heat energy closure or leakage ability of each region of the fabric, for example, the region with high energy constraint function value indicates that the heat energy is easy to accumulate to form local overheating, and the region with low value indicates that the heat is easy to diffuse along the fiber or the fabric gap, and the energy release is fast;

[0078] An energy constraint mapping model inside the material is formed by the global distribution of the energy constraint function; the energy constraint functions of all the equipotential energy regions are integrated to generate a global energy constraint mapping model covering the entire fabric sample, which not only reflects the accumulation and leakage state of the local heat energy, but also reveals the heat flow distribution structure of the whole fabric, through the mapping model, the high-risk heat energy accumulation area, the slow energy conduction area, and the possible local overheating area are identified.

[0079] The process of deriving the time bias distribution of energy conduction in S3 is as follows:

[0080] The time sequence layering processing is performed on the energy constraint mapping model to extract the relative conduction time delay of each energy channel; the energy constraint function of each fabric region corresponds to the accumulation and release characteristics of the local heat in the generated energy constraint mapping model, the mapping model is layering processed along the time axis to form continuous time sequence data, each layer represents the thermal potential change in a specific time window, the energy conduction process of each local region is regarded as an independent energy channel, the relative conduction time delay information between different energy channels is extracted by comparing the starting point and the peak value appearance time of the thermal potential change in the continuous time layers, so as to depict the sequence and delay characteristics of the heat transfer inside the fabric;

[0081] The time offset correlation structure is constructed by analyzing the time delay offset rate between energy channels through a sliding regression window; after obtaining the relative time delay information of each energy channel, the data is continuously scanned and analyzed by using the sliding regression window method, the conduction time delay of multiple energy channels is regression fitted in each time window, the change trend and offset rate are calculated to capture the possible local lag or acceleration phenomenon in heat transfer, the time delay change curve of the entire fabric sample is obtained through window-by-window rolling analysis, and the curves are integrated into the time offset correlation structure, which can reflect the synchronization, offset degree and potential non-uniform conduction mode between different regional energy channels;

[0082] In combination with the fabric thickness, thermal conductivity and surface roughness, the time offset correlation structure is balanced and corrected in parameters to output the time offset distribution; after constructing the preliminary time offset correlation structure, the physical properties of the fabric need to be modified, the influence of the thickness of the fabric on the lengthening of the heat diffusion path, the adjustment effect of the thermal conductivity of the fiber material on the heat flow rate, and the local interference of the surface roughness on the heat exchange efficiency are considered, the time delay of each channel in the time offset correlation structure is balanced and adjusted in parameters, so that it is more in line with the actual physical conditions, thereby eliminating abnormal deviations caused by thickness difference or fiber characteristics, and the output time offset distribution not only reflects the relative conduction time sequence of heat in the fabric, but also provides reliable spatial-time mapping information.

[0083] S4: introducing a spectral equalization function under the time offset distribution, performing asymmetric compensation on the local abnormal points of the infrared response data, constructing a fabric combustion stability indicator factor by using multi-dimensional disturbance residuals, and obtaining a dynamic trajectory of the demarcation threshold between the stable region and the unstable region.

[0084] The process of performing asymmetric compensation on the local abnormal points of the infrared response data in S4 is as follows:

[0085] The infrared response data is unfolded into a spectral energy sequence under the constraint of the time offset distribution; the infrared response data of each observation point on the surface of the fabric is rearranged according to the constraint of the aforementioned time offset distribution, so that the data sequence of each collection point can truly reflect the time characteristics and lag effect of local heat conduction, the frequency analysis is performed on the rearranged time sequence signal, the infrared response of each observation point is divided according to different frequency bands to form a spectral energy sequence, each sequence records the energy change of the corresponding frequency band within the entire collection period, not only the time sequence characteristics of the original thermal behavior are retained, but also the local rapid change and long-term trend are decomposed into different frequency bands;

[0086] According to the local variance of the energy sequence, the abnormal point region is determined, and the adaptive asymmetric compensation function is used to correct the abnormal point; the local variance of each sequence is calculated by using the sliding window statistical method, so as to evaluate the energy fluctuation intensity of each time segment, and the region with a local variance significantly deviating from the mean value of the neighborhood is determined as a potential abnormal point region, which may correspond to the local overheating, heat backflow or heat surge phenomenon of the fabric. Through the continuity analysis of the variance change, the duration and fluctuation amplitude of the abnormal event can also be judged. In order to avoid the influence of single noise point on the judgment, the spatial neighborhood information is combined in the abnormal judgment process, the adjacent pixel abnormal regions are aggregated, and a more reliable abnormal region range is formed; for the identified abnormal point region, an adaptive asymmetric compensation function is designed, which is directionally corrected according to the heat change trend before and after the abnormal point, instead of simple average or symmetric smoothing. For the surge abnormality, the compensation function will gradually adjust the energy level of the stable region before and after the abnormal point, so that the energy value of the abnormal point is smoothly transitioned to a reasonable level. For the sudden decrease of abnormality, the local energy response is dynamically enhanced to restore its thermal behavior characteristics. The compensation process considers the time bias distribution at the same time, ensuring that the adjusted thermal response data not only conforms to the local heat conduction law, but also maintains the global consistency of the whole thermal field, avoiding the destruction of the original thermal mode;

[0087] A multi-dimensional disturbance residual is generated by using a time sequence residual vector, and a disturbance main feature is extracted by principal component analysis, so as to construct a combustion stability indicator according to the disturbance main feature; after the compensation is completed, the difference between the original infrared response sequence and the compensated sequence of each observation point is calculated as a time sequence residual vector, which directly reflects the effect of local thermal disturbance and compensation adjustment. The residual vector of all observation points is collected to form a multi-dimensional disturbance residual field, so as to describe the spatial distribution and time evolution characteristics at the same time. This step provides a rich information basis for extracting the main features of the fabric combustion stability, so that the influence of local abnormal points on the overall analysis can be quantified, tracked and visualized. The principal component analysis is performed on the multi-dimensional disturbance residual, and the principal component feature which best represents the local thermal disturbance mode of the fabric is extracted. These principal component features reflect the most significant thermal instability change of the fabric under combustion or high temperature response. According to the time sequence distribution and amplitude information of the principal component feature, a combustion stability indicator is constructed, which maps the complex multi-dimensional disturbance information into an easily understood index.

[0088] The process of obtaining the dynamic trajectory of the threshold of the stable region and the unstable region in S4 is as follows:

[0089] The time series change of the combustion stability indicator is subjected to cluster segmentation, and a characteristic mutation point and a corresponding energy threshold are calibrated; the change of the combustion stability indicator in the entire time series is continuously collected and recorded to form a complete time series curve, the cluster segmentation method is used to analyze the time series data, time periods with similar change patterns are classified into the same class, thereby identifying different stages of energy response, and by comparing the statistical characteristics of each class, a significant mutation point in the change of the indicator is calibrated, and a corresponding energy threshold is assigned to each mutation point;

[0090] On the basis of the clustering results, the energy distribution difference between the stable region and the unstable region is calculated to generate a dynamic evolution curve reflecting the transition law of thermal response; according to the stable and unstable regions obtained by cluster segmentation, the energy distribution characteristics in each region are extracted, including the average energy, fluctuation amplitude and energy flow concentration degree, the energy distribution difference between adjacent stable and unstable regions is calculated to quantify the transition amplitude and change trend of thermal response between regions, and the energy difference at each time point is continuously plotted to form a dynamic evolution curve, thereby directly presenting the evolution law and transition rate of thermal response from stable to unstable;

[0091] A sliding window integral algorithm is used to track the trend of the dynamic evolution curve and perform dynamic weight balancing to form a dynamic trajectory of the stable region and the unstable region; after obtaining the dynamic evolution curve, local integral analysis is performed on the curve by setting a sliding window to capture the energy change trend and local peak position, and a dynamic weight balancing method is used to adjust the influence weight of energy change in different time periods, so that the window integral result can more accurately reflect the trend of local energy rise or fall, and the trend change and weight correction result are combined to draw a dynamic trajectory of the stable region and the unstable region, and the trajectory evolves with time to show the local combustion stability change of the fabric.

[0092] S5: Based on the decomposition threshold dynamic trajectory, the flame retardant response balance rate and the energy flow retention rate are calculated to form a flame retardant performance evaluation index, and by comparing the index fluctuation range under different thermal excitation levels, the flame retardant performance of the fabric is quantitatively graded and determined.

[0093] The process of forming the flame retardant performance evaluation index in S5 is:

[0094] The average energy flow and fluctuation amplitude of the stable region and the unstable region are extracted in the demarcation threshold dynamic trajectory; according to the demarcation threshold dynamic trajectory of the stable region and the unstable region, the thermal response of the fabric during combustion is divided into stable and unstable sections, for the stable section, the average value of the heat flow with time is calculated to reflect the ability of the fabric to maintain uniform heat distribution under thermal excitation disturbance, and the fluctuation amplitude of the heat flow in the section is analyzed to measure the small disturbance and non-uniformity of local heat, and for the unstable section, the range and rate of energy flow change are calculated to reveal the starting point and diffusion characteristics of thermal instability;

[0095] The ratio of the energy flow in the steady state region to the total energy input is calculated to obtain the flame retardant response balance rate. The average energy flow in the steady state region is compared with the total heat energy input received by the fabric, and the proportion of the total energy input is calculated, which is defined as the flame retardant response balance rate, which is used to quantitatively reflect the ability of the fabric to maintain local thermal stability during the combustion process. Through this index, the uniformity of heat distribution and energy consumption efficiency in different regions of the fabric under continuous heat disturbance are evaluated, so as to judge the overall flame retardant performance. Not only the absolute size of heat energy is considered, but also the comprehensive ability of the fabric in heat energy dispersion and control is revealed through the proportional relationship;

[0096] The ratio of the energy flow in the steady state region to the total energy input is calculated to obtain the flame retardant response balance rate. The average energy flow in the steady state region is compared with the total heat energy input received by the fabric, and the proportion of the total energy input is calculated, which is defined as the flame retardant response balance rate, which is used to quantitatively reflect the ability of the fabric to maintain local thermal stability during the combustion process. Through this index, the uniformity of heat distribution and energy consumption efficiency in different regions of the fabric under continuous heat disturbance are evaluated, so as to judge the overall flame retardant performance. Not only the absolute size of heat energy is considered, but also the comprehensive ability of the fabric in heat energy dispersion and control is revealed through the proportional relationship;

[0097] The flame retardant response balance rate and the energy flow retention rate are fused by weighting to generate a comprehensive evaluation index of flame retardant performance. The flame retardant response balance rate and the energy flow retention rate are fused by weighting to form a comprehensive evaluation index of flame retardant performance. According to the importance of the steady state retention ability and the thermal equilibrium ability of different fabric types and application scenarios, the importance of the steady state retention ability and the thermal equilibrium ability of different fabric types and application scenarios are adjusted to ensure that the evaluation index not only reflects the overall thermal stability but also reflects the local instability control ability. The comprehensive evaluation index generated can be used for quantitative comparison, grading determination and optimization design of the flame retardant performance of the fabric, and can also be compared with the test results under different heat excitation levels to realize comprehensive and multi-dimensional evaluation of the flame retardant characteristics of the fabric.

[0098] The process of quantitatively grading and determining the flame retardant performance of the fabric in S5 is as follows:

[0099] According to the comprehensive evaluation index of flame retardant performance, a flame retardant grading model containing multiple threshold values is established. The response results of different fabric samples under experimental conditions are statistically analyzed, the flame retardant performance is divided into several grade intervals by analyzing the distribution characteristics of the comprehensive evaluation index of the samples under multiple heat disturbance, and reasonable threshold ranges are set for each grade. The flame retardant grading model not only considers the difference in the steady state heat flow maintenance ability of the fabric, but also considers the local thermal instability characteristics, so as to ensure that the model can accurately reflect the comprehensive flame retardant ability of the fabric;

[0100] The fluctuation range of the evaluation index under different heat shock levels is matched with the threshold value of the model to obtain the flame-retardant grade of the fabric; in actual measurement, the comprehensive evaluation index of the flame-retardant performance of different fabrics under different heat shock intensity will present a certain fluctuation, in order to accurately determine the fabric grade, the index fluctuation range under each heat shock level needs to be compared with the threshold value in the established flame-retardant grading model one by one, by matching the index value of each sample under each heat shock condition with the grade threshold value, the corresponding flame-retardant grade is determined, and the stability of the index under multiple heat shock levels is statistically analyzed to ensure that the determination result not only considers the performance under single heat shock, but also comprehensively reflects the adaptability of the fabric under different environmental conditions;

[0101] The thermal response mode and energy flow retention characteristics of the samples of each grade are compared and analyzed to verify the distinguishing effectiveness of the model, and the quantitative grading and determination result of the flame-retardant performance is output; after obtaining the flame-retardant grade of the fabric, the thermal response mode and energy flow retention characteristics of the samples in each grade are analyzed and compared in depth, the energy flow distribution, energy flow retention time, local overheating event and other characteristics of the steady-state region and unstable region are statistically and visually analyzed to test whether there is obvious difference between different grades and whether the grading model can effectively distinguish the advantages and disadvantages of the flame-retardant performance of the fabric, the comparative analysis is not only used to verify the scientificity and distinguishing ability of the model, but also can find the potential weaknesses of the fabric under specific heat shock conditions, which provides a reference for fabric improvement design or performance optimization, the analysis results are arranged to form a standardized report, and the quantitative grading and determination result of the flame-retardant performance of the fabric is output.

[0102] It should be noted that the relationship terms such as first and second in this text are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between the entities or operations. Moreover, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0103] Although the embodiments of the present application have been shown and described, it can be understood by those of ordinary skill in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for analyzing the flame retardancy of fabrics based on infrared thermal imaging sensor data, characterized in that, Includes the following steps: Non-contact temperature measurement of fabric samples is performed in a controlled thermal disturbance field to obtain infrared imaging signal sequences. The infrared imaging signal sequences are then rearranged by energy projection to convert the radiant energy of each pixel into a heat flow migration vector, forming a primary thermal behavior field that characterizes the local thermal conductivity response of the fabric. Based on the discontinuous change in energy flux density in the primary thermal behavior field, a boundary perturbation model of the local overheated domain is established. By comparing the morphological differences between the thermal expansion path and the radiation energy decay zone, the thermal transition mode of the combustion initiation zone is identified, and a critical response map reflecting the thermal instability trend is generated. The thermal potential differences in each region of the critical response diagram are mapped to the energy constraint state inside the material. Combined with the fabric structure density and thermal resistance gradient parameters, the time offset distribution of energy conduction is derived. A spectral equalization function is introduced under time-biased distribution to perform asymmetric compensation on local outliers in infrared response data. Multidimensional disturbance residuals are used to construct fabric combustion stability indicator factors to obtain dynamic trajectory of the boundary between steady-state and unstable regions. Based on the dynamic trajectory of the decomposition threshold, the flame retardant response equilibrium rate and energy flow retention rate are calculated to form flame retardant performance evaluation index. By comparing the fluctuation range of the index under different heat shock levels, the flame retardant performance of the fabric is quantitatively graded and judged.

2. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 1, characterized in that, The process of forming a primary thermal behavior field characterizing the local thermal conductivity response of the fabric is as follows: A multi-channel infrared acquisition unit was set up in a controlled thermal disturbance field to perform time-series scanning on the surface of the fabric sample. Energy integration and temperature difference layering are performed on continuously acquired infrared pixel data, and energy abrupt changes between adjacent frames are defined as transient heat conduction events. The local heat flow direction is calculated by the energy gradient vector between pixels, and the heat flow vector is spatially rearranged by combining the fabric texture distribution. Using heat flux vector density and gradient change rate as core parameters, a local thermal conductivity response function domain is constructed to form a primary thermal behavior field characterizing the local thermal conductivity response of the fabric.

3. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 2, characterized in that, The process of establishing the boundary perturbation model for the local overheated domain is as follows: The continuity of the magnitude sequence of heat flux vectors in the primary thermal behavior field is detected, and gradient abrupt change points and their neighborhood distribution are identified. A local energy flow perturbation function is constructed within the neighborhood of the mutation point, and the region of abnormal energy flow accumulation is defined as a potential overheating domain. The geometric boundary of the superheated domain is determined by a region growth algorithm based on topological adjacency, and the boundary perturbation coefficient is calculated. By coupling analysis of the perturbation coefficient and the energy spread rate, a boundary perturbation model characterizing the degree of local thermal imbalance is formed.

4. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 3, characterized in that, The process of generating a critical response diagram that reflects the trend of thermal instability is as follows: Based on the boundary perturbation model, the energy flow extension path is extracted, and the extension direction of each path is matched with the energy decay rate to form a corresponding sequence. A morphological difference matching algorithm was used to compare the spatial consistency between the extended path and the radiative energy decay region, and to screen out the local energy flow inversion region. By statistically analyzing the abrupt change points of thermal expansion rate and the energy flow reversal intervals within the energy flow inversion region, the thermal transition modes in the combustion initiation region are identified. The thermal instability trend surface is generated based on the distribution density of the thermal transition mode, and then superimposed on the primary thermal behavior field to form a critical response map.

5. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 4, characterized in that, The process of mapping the thermal potential differences in different regions of the critical response diagram to the energy confinement state inside the material is as follows: The critical response map is divided into several equipotential energy regions, and the thermal potential distribution is fitted based on the infrared gray-scale equilibrium value and the thermal conductivity gradient. Within each isopotential energy region, tissue density parameters and fiber alignment direction vectors are introduced to perform scale readjustment of the thermal potential difference. An energy constraint function is constructed by using the directional gradient of the thermal potential difference and the energy dissipation rate to reflect the local energy closure or release state of the fabric. An energy constraint mapping model is formed inside the material based on the global distribution of the energy constraint function.

6. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 5, characterized in that, The process of deriving the time offset distribution of energy conduction is as follows: Perform time series hierarchical processing on the energy-constrained mapping model to extract the relative conduction delay of each energy channel; A time offset correlation structure is constructed by analyzing the time delay offset rate between energy channels through a sliding regression window. By combining fabric thickness, thermal conductivity, and surface roughness, the time-biased correlation structure is parametrically balanced and adjusted to output the time-biased distribution.

7. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 6, characterized in that, The process of performing asymmetric compensation on local outliers in infrared response data is as follows: The infrared response data is expanded into a spectral energy sequence under time-biased distribution constraints; Anomaly regions are determined based on the local variance of the energy sequence, and an adaptive asymmetric compensation function is used to correct the anomalies. Multidimensional perturbation residuals are generated using time-series residual vectors, and principal component analysis is used to extract the main perturbation features to construct combustion stability indicator factors.

8. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 7, characterized in that, The process of obtaining the dynamic trajectory of the boundary between the steady-state region and the unstable region is as follows: Clustering and segmentation are performed on the time-series changes of combustion stability indicator factors to identify characteristic abrupt change points and corresponding energy thresholds; Based on the clustering results, the energy distribution difference between the steady-state region and the unstable region is calculated to generate a dynamic evolution curve that reflects the transition law of thermal response; A sliding window integral algorithm is used to perform trend tracking and dynamic weight balancing on the dynamic evolution curve, forming a dynamic trajectory that forms the boundary between the steady-state region and the unstable region.

9. The fabric flame retardancy analysis method based on infrared thermal imaging sensor data according to claim 8, characterized in that, The process of formulating flame retardant performance evaluation indicators is as follows: Extract the mean energy flow and fluctuation amplitude of the steady-state and unstable regions from the dynamic trajectory of the boundary threshold; The flame retardant response equilibrium rate is obtained by calculating the ratio of steady-state energy flow to overall energy input. The energy flow retention rate is calculated as the ratio of the energy decay rate in the unstable region to the energy retention time in the steady-state region. The flame retardant response equilibrium rate and energy flow retention rate are weighted and fused to generate a comprehensive evaluation index for flame retardant performance.

10. The method for analyzing the flame retardancy of fabrics based on infrared thermal imaging sensor data according to claim 9, characterized in that, The process of quantitatively classifying and determining the flame retardant properties of fabrics is as follows: Based on the comprehensive evaluation index of flame retardant performance, a flame retardant classification model including multiple threshold levels is established. The flame retardant rating of the fabric is obtained by matching the fluctuation range of the evaluation index under different heat shock levels with the model threshold. The thermal response modes and energy flow retention characteristics of samples at different levels are compared and analyzed to verify the model's distinguishing effectiveness and output quantitative classification and judgment results of flame retardant performance.

Images