Method and system for testing performance of high-temperature-resistant cable

Abstract

The invention relates to the technical field of electrical digital data processing, and discloses a high-temperature-resistant cable performance test method and system, and the method comprises the steps: synchronously collecting thermal imaging, optical imaging and gas component data flows in a combustion test, and building a shared time-space domain for alignment; when the temperature reaches the glass transition temperature of the material, crack and gassing events are detected in a constrained time sequence window, events passing error checking are paired to form a causal event chain, and then a structured pyrolysis failure fingerprint is generated. The method does not depend on a single static index at the end of the test, but reconstructs and quantifies isolated measurement results of different physical dimensions in the whole combustion process into a dynamic failure process model with internal time sequence and causal association, so that the information density and decision value of the test are improved.

Description

Technical Field

[0001] This invention relates to a method and system for testing the performance of high-temperature resistant cables, belonging to the field of electronic digital data processing technology. Background Technology

[0002] In engineering fields such as aerospace and nuclear energy facilities, where reliability requirements are stringent, the performance evaluation of high-temperature resistant cables is a key link in ensuring system safety. Currently, the industry generally adopts a standardized combustion test, which involves applying a defined heat load and measuring one or more macroscopic, static endpoint indicators, such as flame spread length or whether electrical functions are connected, to determine whether the cable sample is qualified. This method, with its clear results and good repeatability, has become the cornerstone of quality certification and standard compliance testing in the industry.

[0003] However, with the advancement of materials science and the increasing demand for a deeper understanding of failure mechanisms in systems safety engineering, the inherent limitations of the aforementioned evaluation method, which simplifies a complex dynamic combustion process into a single static endpoint, are becoming increasingly apparent. While providing a pass / fail assessment, it also obscures a large amount of crucial information carried by the failure process itself. For example, two cables of the same final burned length may have vastly different risk profiles: one may rapidly release a large amount of corrosive gas in the early stages of combustion, while the other may experience structural collapse in the later stages. For engineers who need to conduct safety design or accident tracing, these two drastically different failure processes correspond to vastly different risk profiles, and existing test data fails to provide such process information.

[0004] To obtain richer information, a seemingly direct approach is to add more types of sensors to the experiment to collect multi-dimensional data. However, this simple physical superposition does not address the core issue. The result is merely several data curves that are parallel in time but still physically isolated, failing to resolve a deeper contradiction. Specifically, existing technologies have the following shortcomings: 1. The lack of a unified spatiotemporal benchmark among multi-source heterogeneous data makes the correlation analysis between different physical phenomena lack objective basis; 2. The data processing logic remains passive in recording and presentation, lacking a mechanism to actively identify and extract causal relationships between different data streams; 3. The final output is still a discrete set of observation results, rather than a structured process model that can completely reproduce the process from the initial trigger to the final failure and has an inherent logical relationship. Therefore, how to design a data processing method that can go beyond simple data acquisition, and through in-depth processing of multimodal heterogeneous data streams acquired synchronously during combustion, automatically identify and construct the temporal and causal relationships between key physical events, thereby transforming an invisible dynamic failure process into an analyzable structured data object, becomes the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides a method and system for testing the performance of high-temperature resistant cables. Its main purpose is to solve the problem that existing testing methods can only provide static endpoint results and cannot quantitatively characterize and trace the causes of dynamic failure processes of materials under thermal stress.

[0006] To achieve the above objectives, the present invention provides a method for testing the performance of high-temperature resistant cables, the method comprising: During the continuous combustion test on the cable sample, thermal imaging data stream, optical imaging data stream and gas composition data stream were acquired synchronously with the same master clock. Based on the optical imaging data stream, the axial physical coordinates of the cable under test are calibrated, and the thermal imaging data stream is registered to these axial physical coordinates to establish a shared spatiotemporal domain. Within this shared spatiotemporal domain, when the temperature at a certain location along the axial physical coordinate first reaches the glass transition temperature of the cable material... If the time is right, then the first detection window and the second detection window will be opened respectively, based on that time. Within the first detection window, crack events are detected from the optical imaging data stream based on pixel gradient change rules, and within the second detection window, the concentration inflection point of the target component is detected from the gas composition data stream. Before performing timing pairing, the synchronization and registration errors of the thermal imaging data stream, optical imaging data stream and gas composition data stream are checked, and data frames with errors exceeding the given tolerance are discarded. Only the temperature rise, cracking event and concentration inflection point verified by error closed loop are time-series paired to form a causal event chain; Based on the formed causal event chain, a pyrolysis failure fingerprint characterizing the dynamic failure characteristics of the cable is generated and output.

[0007] Preferably, the step of calibrating the axial physical coordinates of the cable under test based on the optical imaging data stream specifically includes: using edge detection and feature recognition algorithms to process continuous image frames in the optical imaging data stream to identify and extract the outline of the cable under test in the image; determining the axis passing through the geometric center of the outline based on the outline of the cable under test, and defining the axis as the axial physical coordinate; and establishing a transformation relationship that maps the coordinates of each pixel in the thermal imaging data stream to the axial physical coordinate to perform registration.

[0008] Preferably, the step of detecting the concentration inflection point includes: [the steps are described in the original text, but the provided text is incomplete and cannot be translated accurately.] The process is performed to calculate the concentration value. First time derivative With the second time derivative Specifically, when the first-order time derivative and the second-order time derivative simultaneously satisfy the following conditions within the second detection window: and When the concentration inflection point is determined to have occurred, then... is a given positive threshold number used to characterize the effective growth rate of concentration.

[0009] Preferably, the method further includes: synchronously acquiring an acoustic emission data stream, which characterizes the high-frequency stress waves emitted by the cable due to internal microscopic fractures under thermal stress; and incorporating acoustic bursting events identified from the acoustic emission data stream as internal mechanical damage events into the pairing and formation process of the causal event chain, so that the final pyrolysis failure fingerprint contains event information that distinguishes between external morphological changes and internal structural damage.

[0010] Preferably, the method further includes: storing multiple generated pyrolysis failure fingerprints into a failure mode database, and using the failure mode database to train a time-series prediction model; and in a new test process, inputting causal event chain fragments generated in real time in the early stage of the test into the time-series prediction model, and having the time-series prediction model output the prediction results of the final failure mode and failure time of the cable.

[0011] Preferably, the duration of the second detection window is set based on the physical distance between the surface of the cable under test and the gas composition data stream acquisition point, as well as the physical delay time determined by the gas transmission rate of the target component under the test environment; the synchronous acquisition frequency is not less than 10 Hz; and the given tolerances are specifically: the synchronization time jitter error is less than one sampling period, and the registration spatial positioning error is less than one pixel size.

[0012] Preferably, the pyrolysis failure fingerprint is a structured data object. Within this structured data object, each event constituting a causal event chain records a unique event type identifier, an absolute timestamp of the event occurrence, the axial physical coordinates of the event occurrence, and the local temperature value and target component concentration value corresponding to the event occurrence.

[0013] Preferably, the step of detecting crack events from the optical imaging data stream based on pixel gradient change rules is as follows: within the first detection window, the pixel gray-level gradient distribution along the radial direction of the cable under test in the optical imaging data stream is calculated; when a local peak appears in the pixel gray-level gradient distribution, and the value of the local peak exceeds the given crack judgment gradient threshold, the crack event is determined to have occurred at that location.

[0014] Preferably, all steps of the method are performed under the constraints of standard procedures and physical conditions that do not interrupt continuous combustion testing, so as to ensure the direct correlation between the pyrolysis failure fingerprint and the existing industry standard testing environment.

[0015] A high-temperature resistant cable performance testing system, the system comprising: A sensing and acquisition module, comprising a thermal imaging sensor, an optical camera, and a gas analyzer, is configured to synchronously acquire thermal imaging data streams, optical imaging data streams, and gas composition data streams at the same master clock during the continuous combustion test of a cable sample. A data processing module is connected to the sensing and acquisition module. The data processing module is configured to: calibrate the axial physical coordinates of the cable under test based on the optical imaging data stream, and register the thermal imaging data stream to these axial physical coordinates to establish a shared spatiotemporal domain; within this shared spatiotemporal domain, when the temperature at a certain location along the axial physical coordinate first reaches the glass transition temperature of the cable material... At that time, the first detection window and the second detection window are opened respectively, based on that time. Crack events are detected from the optical imaging data stream only within the first detection window according to the pixel gradient change rule, and the concentration inflection point of the target component is detected from the gas composition data stream only within the second detection window. Before performing time-series pairing, the synchronization and registration errors of the thermal imaging data stream, the optical imaging data stream, and the gas composition data stream are checked, and data frames with errors exceeding the given tolerance are discarded. Only the temperature rise, crack events, and concentration inflection points that pass the error closed-loop check are time-series paired to form a causal event chain. Based on the formed causal event chain, a pyrolysis failure fingerprint characterizing the dynamic failure characteristics of the cable is generated and output.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This method simultaneously collects at least two heterogeneous data streams from different physical dimensions during combustion tests. Under a unified spatiotemporal reference, it identifies and extracts key features from the data streams based on preset causal association rules. This integrates multiple measurement results that were originally isolated on the time axis and physically fragmented into a failure event chain with inherent temporal and logical relationships. This approach makes the evaluation of cable performance no longer dependent on a static endpoint indicator at the end of the test, but rather on the objective reproduction of the dynamic failure process composed of multiple domain events throughout the combustion process. The evaluation results generated and output in this way have transformed from one-dimensional numerical values ​​or states into a high-dimensional data object that can depict the evolution path of failure modes.

[0017] 2. When the acquired data stream further includes acoustic emission data stream, the data processing module will correlate the high-frequency stress wave events generated by micro fractures inside the material with hotspot events in the thermal imaging data stream and morphological change events in the optical imaging data stream in time. In this way, the evaluation dimension is extended from the external characterization of the cable to its internal mechanical integrity evolution. This makes the failure event chain not only record visible surface phenomena, but also include the invisible internal damage accumulation process. Thus, it is possible to make an objective and quantitative distinction between burst failure and progressive failure, two failure modes that are completely different in mechanism but may have similar macroscopic results.

[0018] 3. The structured evaluation results generated by this method, which characterize the dynamic failure properties of cables, can be included in a failure mode database because they record a continuous event chain from initial heating to final failure. When a deep learning model is trained using this database, it can analyze the event chain segments being generated in real time and compare them with existing patterns in the database during a new test. In the early stages of the test, the final failure mode and approximate failure time of the cable can be predicted. This shift in testing and analysis methods allows the focus of testing to shift from obtaining final conclusions to understanding early trends, providing a time advantage for the research and development iteration process of new materials. Attached Figure Description

[0019] Figure 1 This is a flow chart of a high-temperature resistant cable performance testing method according to the present invention; Figure 2 This is a schematic diagram illustrating the determination of the critical temperature rise event in this invention. Figure 3 This is a functional architecture block diagram of a high-temperature resistant cable performance testing system according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] The present invention provides a method and system for testing the performance of high-temperature resistant cables. The data processing flow consists of the following stages: First, during the continuous combustion test of the cable sample, a sensing and acquisition module synchronously acquires data streams from at least three different physical dimensions using the same master clock. Subsequently, in a data processing module, a shared spatiotemporal domain is established for these data streams from different sources through coordinate transformation and registration algorithms. Finally, within this unified spatiotemporal domain, based on a set of causal association rules characterizing physical failure mechanisms, key features in the data streams are identified, matched under time constraints, and error checked. This reconstructs the discrete observation data into one or more failure event chains with inherent logical relationships. Based on these failure event chains, a pyrolysis failure fingerprint characterizing the dynamic failure characteristics of the cable is generated and output.

[0022] In high-temperature cable combustion tests, the cable failure process is the result of the spatiotemporal coupling of multiple physical fields such as heat, force, and chemistry. If the consistency of data streams from various sensors on the time reference cannot be ensured at the source of data acquisition, subsequent data processing aimed at exploring causal relationships will lack an objective basis. To address this, the present invention synchronously acquires thermal imaging data streams, optical imaging data streams, and gas composition data streams using the same master clock during the continuous combustion test of the cable sample. Specifically, a central clock source generates a frequency-stable signal, which is distributed to the thermal imaging sensor and the optical camera. Each of the data acquisition units of the gas analyzer serves as the sole sampling trigger reference. The system is set to synchronously acquire data at a frequency of no less than 10 Hz. This frequency is set because it is higher than the characteristic frequencies of physical events such as microcrack propagation and instantaneous thermal decomposition of materials during cable combustion, thus meeting the basic requirements for signal reconstruction. In each sampling cycle, each data frame or data point acquired by each sensor is appended with a unique timestamp generated by the count value of the central clock source. In this way, the consistency and traceability of the time reference required for subsequent spatiotemporal alignment and causal correlation analysis are ensured at the data level.

[0023] Furthermore, even though the data streams are synchronized in time, they still exist in independent coordinate systems defined by different sensors. For example, the thermal imaging data stream exists in the pixel coordinate system of the infrared focal plane array, while the optical imaging data stream exists in the pixel coordinate system of the visible light sensor. This difference in spatial reference makes it difficult to directly correlate the temperature rise at a certain location with the morphological change at the same physical location. In view of this, the system adopts the following procedure: based on the optical imaging data stream, the axial physical coordinates of the cable under test are calibrated, and the thermal imaging data stream is registered to these axial physical coordinates to establish a shared spatiotemporal domain. The execution of this procedure includes two steps. First, before the combustion test begins, an edge detection and feature recognition algorithm is used to process a frame of static image in the optical imaging data stream to identify and extract the outline of the cable under test in the image. Based on the geometric center of the outline, the axis passing through the outline is calculated and defined as a one-dimensional axial physical coordinate. Simultaneously, a pixel coordinate system is established that maps the optical image along this axis. Mapping to physical coordinates Transformation relationship Secondly, by using a known physical location of the cable sample... A controllable heat source is applied at a specific point, and thermal and optical imaging data are acquired simultaneously. The data processing module identifies the pixel coordinates of the hot spot in the thermal image. and pixel coordinates in the optical imaging map By establishing multiple sets of such corresponding point pairs, an affine transformation matrix is ​​fitted. This matrix defines the mapping of the coordinates of each pixel in the thermal imaging data stream to the optical imaging coordinate system, and finally transforms the coordinates through the transformation relationship. The transformation path, mapped to axial physical coordinates, is used to perform this registration procedure, and the system then maps all subsequently acquired data... All data streams are unified into a shared spatiotemporal domain based on physical time and the physical coordinates of the cable axis. This provides a unified reference system for subsequent event identification.

[0024] After establishing a shared spatiotemporal domain, the task of data processing is to identify discrete events that indicate changes in the material state from a continuous data stream and to construct causal relationships between them. To correlate changes in the material's physical properties, such as the glass transition, with their subsequent mechanical property degradation and chemical decomposition, the present invention is configured to, within this shared spatiotemporal domain, at a certain position along the axial physical coordinate... temperature The glass transition temperature of this cable material was reached for the first time. Then, at that time... Based on this, the first and second detection windows are opened respectively; the glass transition temperature here... These are the inherent property values ​​of the cable sample material obtained through offline physical performance testing using differential scanning calorimetry; As the starting point for event chain identification, the algorithm's initiation is based on the physical fact of material state transitions. Once the temperature rise trigger event is confirmed, the system enters a time-constrained search phase targeting its physical consequences, within the first detection window. Within the second detection window, crack events are detected from the optical imaging data stream based on pixel gradient change rules. Within, the concentration inflection point of the target component is detected from the gas composition data stream; among which, the duration of the first detection window... It is determined based on the thermodynamic response time of the material. Within this window, the algorithm calculates the optical image along the radial direction of the cable under test. The pixel grayscale gradient distribution at a given location; when a local peak appears in this distribution and the value of the peak exceeds a preset crack detection gradient threshold. When the time is right, a crack event is determined to have occurred; however, the opening time of the second detection window includes a physical delay. The procedure for determining this time is as follows: ,in, This refers to the physical distance between the surface of the cable under test and the sampling probe of the gas analyzer. The average gas flow rate within the test chamber under the combustion conditions is both a system physical parameter; the detection of the concentration inflection point within the second detection window is achieved by calculating the target component concentration value. First time derivative With the second time derivative This is achieved when the following conditions are met simultaneously within the window: and When the concentration inflection point is determined to have occurred, then... The positive threshold for characterizing the effective growth rate of concentration is set based on a value higher than the baseline noise level of the gas analyzer in a static environment.

[0025] To ensure the objectivity of the final causal event chain, before time-series pairing, the synchronization and registration errors of the thermal imaging data stream, optical imaging data stream, and gas composition data stream are checked, and data frames with errors exceeding a given tolerance are discarded. The given tolerance is set as follows: synchronization time jitter error is less than one sampling period, and registration spatial positioning error is less than one pixel size. Only temperature rise, crack events, and concentration inflection points that pass this error closed-loop check are allowed for time-series pairing to form a causal event chain. For example: [Event 1:] → [Event 2: Crack ( Event 3: Concentration Inflection Point )],in, and Finally, based on the formed causal event chain, the system generates and outputs a pyrolytic failure fingerprint characterizing the dynamic failure characteristics of the cable. This pyrolytic failure fingerprint is defined as a structured data object. Within this object, for each event constituting each causal event chain, a unique event type identifier, an absolute timestamp of the event occurrence, the axial physical coordinates of the event occurrence, and the corresponding local temperature value and target component concentration value at the time of the event occurrence are recorded. This structured output transforms a dynamic failure process into a quantifiable, analyzable, and comparable digital object to support the performance evaluation and failure mechanism analysis of high-temperature resistant cables.

[0026] Example 1: The technical solution of the present invention operates as follows in a selection and verification scenario for an aerospace component: In this scenario, a high-temperature resistant cable inside the servo control system of a new launch vehicle needs to pass through the high-temperature region of the engine. The engineering team faces the decision of selecting the best among two candidate cable samples, namely Sample A and Sample B. Both samples exhibit consistent performance in traditional combustion tests, specifically the same flame spread length and the same electrical function failure time. This result makes it impossible to make risk differentiation and selection judgment based on the test data. To address this decision dilemma, the method described above is used to perform a performance test on Sample A and Sample B respectively. During the test, the sensing and acquisition modules synchronously acquire thermal imaging data streams, optical imaging data streams, and gas composition data streams using the same master clock. The data processing module then calibrates the axial physical coordinates of the tested cable based on the optical imaging data stream and completes the registration of the thermal imaging data stream, thereby establishing a shared spatiotemporal domain. For Sample A, the data processing module first identifies a section located in the middle of the cable within this shared spatiotemporal domain. The temperature rise event was the first to reach the glass transition temperature of the material. Within the detection window opened based on this, the system detected a hydrogen chloride gas concentration inflection point in the gas composition data stream after a set time delay. However, no crack event conforming to the pixel gradient change rule was detected in the corresponding optical imaging data stream. For sample B, the data processing module detected the crack near the cable end at a relatively higher temperature. Temperature rise event detected In the first detection window, multiple axially extending crack events were continuously detected from the optical imaging data stream, but in the corresponding second detection window, the hydrogen chloride concentration in the gas composition data stream did not show a clear concentration inflection point.

[0027] After completing the error closed-loop verification of the test data of the two samples, the data processing module performed time-series pairing on them and generated pyrolysis failure fingerprints with structural differences. The core causal event chain of sample A's pyrolysis failure fingerprint was recorded as a direct triggering of chemical decomposition events by a critical temperature rise event. That is, after the material reaches its glass transition temperature, its main failure path is characterized by the breaking of chemical bonds and the release of corrosive gases, while the overall integrity of its physical structure is maintained for a relatively long period. In contrast, sample B's pyrolysis failure fingerprint depicts a different failure path. Its causal event chain is recorded as a series of chain-like mechanical failure events triggered by a critical temperature rise event. That is, the material reaches the point of mechanical property degradation under thermal stress, and its main failure path is characterized by the initiation and propagation of microcracks, ultimately leading to the structural collapse of the cable. This discriminative capability stems from the fact that synchronous acquisition ensures that different materials... The temporal correlation of events was analyzed, and spatiotemporal alignment provided a unified spatial benchmark for this correlation. This shifted the focus of evaluation from the identical failure results of the two samples to the differentiation of their underlying failure mechanisms and evolution paths. The original question of which cable was more durable was transformed into the analyzable question of which failure mode of the two cables had a more controllable risk. Ultimately, the engineering team made a selection decision based on two different pyrolysis failure fingerprints. Although the failure mode of sample A would release corrosive gases, its timing and location could be quantified and recorded by the test method, so its risk could be managed by isolating and protecting surrounding devices. On the other hand, the failure mode of sample B was a sudden structural collapse, which was more unpredictable and posed a direct risk to the functional safety of the servo control system. Therefore, sample A was selected as the technical solution.

[0028] Example 2: To objectively verify the effectiveness of the method of the present invention in distinguishing different failure modes, this example designed and executed a comparative experiment. The purpose of the experiment was to quantitatively compare the differences in test results obtained from two high-temperature resistant cable samples with different failure mechanisms but consistent macroscopic thermal properties when using the traditional endpoint evaluation method and the method of the present invention. The test platform was built in a combustion test chamber conforming to the IEC60332-1 standard, which deployed the sensing and acquisition modules defined in the aforementioned specific embodiments, including an uncooled vanadium oxide thermal imaging sensor, an optical camera with a resolution of 1920x1080 pixels, and a Fourier transform infrared spectroscopy gas analyzer. All sensor data acquisition was synchronized and timestamped by a single data processing module using the same master clock at a frequency of 10 Hz. Two samples were used in this experiment: a control sample and the present invention sample. The control sample used chlorine-rich polyvinyl chloride as insulation, whose failure path under thermal stress tends towards chemical decomposition. The present invention sample used a modified cross-linked polyolefin insulation material, which becomes more brittle after reaching its glass transition temperature, and its failure path tends towards microcrack propagation caused by thermal stress. The conductor specifications, insulation thickness, and rated temperature resistance were consistent for both samples. Before the experiment, parameters related to material properties, such as the glass transition temperature, were entered into the data processing module. The settings were all based on the differential scanning calorimetry test results of the two materials.

[0029] Experimental results show that the method of this invention can identify and quantify the differences in failure mechanisms between the two sample groups, while traditional endpoint testing cannot provide this level of differentiation. Specifically, when using the traditional endpoint testing procedure, the electrical function failure time of the control sample group was recorded as 185 seconds, while that of the sample group of this invention was recorded as 188 seconds, with no difference between the two values ​​within the experimental error range. However, when the complete testing method disclosed in this invention is used, the generated pyrolysis failure fingerprint exhibits differences in structure and event sequence. For the control sample group, the data processing module captures the temperature reaching its material level at 82 seconds of combustion. The value triggers the event, and subsequently, within a second detection window corrected for physical delay time, identifies a chemical decomposition event marked by hydrogen chloride precipitation; in contrast, the sample group of the present invention The triggering event occurred at 85 seconds, immediately followed by a mechanical cracking event identified within the first detection window, characterized by pixel grayscale gradient changes. This data comparison confirms that the single time scalar output by traditional testing methods cannot reveal the differences in the failure processes of the two sample groups. In contrast, the method of this invention, through causal correlation processing based on physical mechanisms of the multimodal data stream, outputs a pyrolysis failure fingerprint that not only records the similarities between the two sample groups... The trigger time was further distinguished from the two different failure paths in the form of a structured causal event chain: one was dominated by chemical decomposition and the other by mechanical cracking. The test results confirmed that the data processing method of the present invention can extract and construct process information about the evolution of failure modes from the same physical process as traditional testing, thereby providing an analytical basis based on failure mechanism for cable performance evaluation.

[0030] Example 3: This example combines Figures 1 to 3 A description of a test method and system for the performance of a high-temperature resistant cable, such as... Figure 1 As shown in the figure, this diagram depicts the collaborative relationship and data flow process between various logic modules during the test in the form of signal interaction. It begins with the temperature monitoring module on the left. When this module detects that the local temperature of the cable sample has reached the preset glass transition temperature, it sends a trigger signal to the data processing module. After recording the time and location of the temperature rise event, the data processing module successively opens a first detection window and a second detection window containing physical delay to activate the crack detection module and the gas detection module, respectively. The crack detection module identifies crack events by calculating pixel gradient changes and reports the detection results to the data processing module. The gas detection module determines the concentration inflection point by calculating the first and second derivatives of the concentration and also reports its detection results to the data processing module. After receiving these event reports from different physical dimensions, the data processing module chains the temperature rise event, crack event, and concentration inflection point into the causal event chain module, which finally generates a structured pyrolysis failure fingerprint.

[0031] like Figure 2 As shown in the figure, the horizontal axis represents time in seconds (s), and the vertical axis represents temperature in degrees Celsius (°C). The figure includes a solid line representing the measured temperature curve and a dashed line representing the glass transition temperature of the material. The threshold value was set at 120°C. Over time, the measured temperature curve showed a monotonically increasing trend, and at approximately 130 seconds... When the threshold lines intersect, the intersection point signifies that a valid critical temperature rise event has been triggered, thereby initiating a series of subsequent related event detection and causal chain construction processes.

[0032] like Figure 3As shown in the figure, this diagram depicts the complete processing chain from data acquisition to final output. The object of processing is the cable sample under test. On the left is the sensing and acquisition module. Under the control of a synchronous trigger signal generated by a master clock, this module observes the state of the sample in the combustion test through its integrated thermal imaging sensor, optical camera, gas analyzer, and acoustic emission sensor as a key extension module, and outputs multimodal data streams. These data streams are sent to the core data processing module on the right. This module contains, in sequence, a spatiotemporal alignment unit responsible for unifying the spatiotemporal reference of the data, a key event detection unit responsible for identifying state change points from the data stream, a causal event chain construction unit responsible for constructing event correlations based on physical rules, and finally, a pyrolysis failure fingerprint generation unit responsible for encapsulating the output results. The pyrolysis failure fingerprint generated by this unit is a structured data object. On the one hand, it can be directly used for performance evaluation. On the other hand, it can be stored in a failure mode database, which can then be called to train a time-series prediction model. This model outputs a predictive assessment of the final result in the early stages of a new test.

[0033] Example 4: In a specific application scenario, to ensure that the internal logical parameters of the data processing module of the method of the present invention match the specific experimental physical environment before formal testing, a systematic calibration procedure needs to be executed. This procedure aims to establish a traceable and reproducible data processing benchmark for all subsequent tests. It is executed in a combustion test chamber with sensors and acquisition modules installed and integrated. The core objective is to determine the quantified values ​​of the gas transmission physical delay time, the crack event judgment gradient threshold, and the effective growth rate threshold of the gas concentration inflection point in the data processing module. The procedure first calibrates the gas transmission physical delay time. Given that it takes a certain amount of time for the gas components precipitated from the cable surface to be transmitted to the sampling probe of the gas analyzer, this time delay is a prerequisite for correctly correlating the temperature rise event and the chemical decomposition event. To determine this delay, a pulse injector capable of releasing trace amounts of tracer gas is placed at the position of the standard cable sample in the test chamber. Under the control of the data processing module, a pulse injector is used to release trace amounts of tracer gas. An argon gas pulse lasting 100 milliseconds is injected intermittently, while a gas analyzer is configured to continuously monitor changes in argon gas concentration at a frequency of 10 Hz. The data processing module records the moment when the concentration reaches its peak. Furthermore, the average gas flow rate inside the test chamber pass The calculation shows that, among which For the pre-measured physical distance from the injection point to the sampling probe, in this calibration, when It was measured to be 0.5 meters. At 4.8 seconds, The velocity was determined to be approximately 0.104 m / s. This value was stored as a system parameter, and the data processing module will use this velocity to calculate the physical delay time required for the gas released at different locations to reach the sampling probe during subsequent formal testing. .

[0034] Subsequently, the procedure establishes a gradient threshold for determining crack events. Calibration is performed, as the threshold setting is crucial to the sensitivity and reliability of the algorithm in identifying microscopic mechanical damage on the material surface. Therefore, a reference sample with the same insulation material and surface characteristics as the cable under test is used. A series of microscopic damage points with different indentation sizes are pre-fabricated on its surface using a Vickers hardness tester. These sizes are measured and recorded using a microscope, covering a range from 10 micrometers to 50 micrometers. The reference sample is placed in a test chamber, and under standard lighting conditions, its surface image is acquired by an optical camera. The data processing module calculates the pixel grayscale gradient distribution along the radial direction of each pre-fabricated damage point in the image and records the gradient peak corresponding to a predetermined damage size, i.e., 20 micrometers. Finally, the crack judgment gradient threshold is determined. The gradient peak value is set to the sum of three times the standard deviation of the image background noise gradient measured by the system in the absence of a sample. This setting ensures the ability to detect microcracks of critical dimensions while avoiding misjudgments that may be caused by the noise of the image sensor itself.

[0035] Ultimately, the procedure sets an effective growth rate threshold for the inflection point of gas concentration. Calibration is performed, and this threshold is crucial for distinguishing the real gas evolution signal from the analyzer's background noise. The calibration process requires starting the heating and ventilation systems of the combustion test chamber without any sample, ensuring its operation matches that of a real test. Under these conditions, the gas analyzer continuously collects background concentration data of the target gas, namely hydrogen chloride, for 300 seconds. The data processing module receives the data stream and calculates its first-order time derivative. Standard deviation This standard deviation reflects the random fluctuation amplitude of the concentration signal under this working environment, and the effective growth rate threshold. Set as This setting corresponds to a predetermined low false alarm rate statistical confidence level; after this calibration process is completed, the core algorithm parameters of the data processing module are all assigned values ​​derived from physical measurements, thus providing a repeatable basis for the output results of the entire test system.

[0036] Example 5: In a specific application scenario, to enable the failure mode predictive identification function included in the method of the present invention, an offline construction and filling procedure for a failure mode database needs to be executed in advance. This procedure selects multiple sets of reference cable samples with clear material formulations and known failure modes, and conducts a complete combustion test on each of them using the method of the present invention. A corresponding pyrolysis failure fingerprint is generated for each reference sample. Subsequently, the data processing module further performs feature extraction operations on each pyrolysis failure fingerprint in the database, transforming the original causal event chain composed of discrete events into a standardized multidimensional feature vector containing the following dimensions: the density of mechanical events within a preset time window, the average time delay between chemical and thermal events, and the cumulative energy of acoustic explosion events. Finally, these feature vectors and their corresponding known failure mode labels are stored as a data pair in the failure mode database, providing structured and labeled source data for subsequent training of time series prediction models.

[0037] Furthermore, when the test system of the present invention further integrates the acquisition of acoustic emission data stream, in order to avoid the interference of background mechanical vibration in the test environment on the internal micro fracture signal, a pre-calibration procedure for the on-site deployment of the acoustic channel must be performed before conducting a series of test tasks. This procedure requires that, without placing any cable samples, all auxiliary systems such as ventilation and heating of the combustion test chamber be turned on, and the background noise signal be continuously acquired by the acoustic emission sensor for at least five minutes. The data processing module performs spectral analysis on the signal to identify the main frequency characteristics and amplitude distribution of the background noise, and generates a dynamic noise filter accordingly. In the subsequent formal test, the filter is used to preprocess the real-time acquired acoustic emission data stream. A signal is only judged as a valid acoustic burst event and is included in the pairing and formation process of the causal event chain when it simultaneously meets the two conditions that its energy amplitude exceeds five standard deviations of the statistical mean of the background noise and the correlation between its main frequency component and the peak value of the background noise spectrum is lower than a predetermined threshold.

[0038] Example 6: In a specific application scenario, to establish a high-confidence failure mode time series prediction model that can serve as a benchmark for subsequent online prediction, a standardized offline construction procedure including data preprocessing, model training, and performance verification needs to be executed. The basis of this procedure is to ensure that the source data used for model training, namely the thermal failure fingerprint set, has a high degree of integrity and logical consistency. To this end, when calling data from the failure mode database, a preliminary data integrity check is first performed. The data processing module checks the timestamp continuity of the original synchronous data stream frame by frame. If, within a certain time period, any data stream experiences the loss of data frames for more than three consecutive sampling periods, the data within that time period will be marked, and any causal event chain that starts at or spans that time period will be removed from the training dataset.

[0039] Furthermore, to increase the stability of the initial trigger point of the causal event chain, this procedure imposes the following limitations on the judgment logic of critical temperature rise events: A valid temperature rise trigger event is defined as the local temperature value at a certain axial physical coordinate position not only needing to reach the glass transition temperature of the cable material for the first time. Furthermore, at least three sampling cycles must be maintained above this temperature. Only when this condition is met will the data processing module achieve the first sampling. Based on the time frame, the first and second detection windows are opened. After data verification and trigger logic optimization are completed, the feature vectors extracted from the database using the aforementioned procedure are divided into a training set (80% of the total) and a validation set (20%). The time-series prediction model adopts a temporal convolutional network (TCN) architecture, using the training set data to train the model parameters. The training objective is to minimize the cross-entropy loss function between the failure modes predicted by the model and the actual failure modes labeled in the database. The final output of this procedure is a performance-verified time-series prediction model, whose verification standard is that the model's prediction accuracy on the validation set exceeds 95%. This model is deployed to the data processing module, which then has the ability to output the prediction result of the final failure mode of the cable in the early stages of a new test process, based on real-time generated causal event chain fragments that have undergone data integrity verification and stable logic triggering.

[0040] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for testing the performance of high-temperature resistant cables, characterized in that, The method includes: During the continuous combustion test on the cable sample, thermal imaging data stream, optical imaging data stream and gas composition data stream were acquired synchronously with the same master clock. Based on the optical imaging data stream, the axial physical coordinates of the cable under test are calibrated, and the thermal imaging data stream is registered to these axial physical coordinates to establish a shared spatiotemporal domain. Within this shared spatiotemporal domain, when the temperature at a certain location along the axial physical coordinate first reaches the glass transition temperature of the cable material... If the time is right, then the first detection window and the second detection window will be opened respectively, based on that time. Within the first detection window, crack events are detected from the optical imaging data stream based on pixel gradient change rules, and within the second detection window, the concentration inflection point of the target component is detected from the gas composition data stream. Before performing timing pairing, the synchronization and registration errors of the thermal imaging data stream, optical imaging data stream and gas composition data stream are checked, and data frames with errors exceeding the given tolerance are discarded. Only the temperature rise, cracking event and concentration inflection point verified by error closed loop are time-series paired to form a causal event chain; Based on the formed causal event chain, a pyrolysis failure fingerprint characterizing the dynamic failure characteristics of the cable is generated and output.

2. The method for testing the performance of high-temperature resistant cables according to claim 1, characterized in that, The specific steps for calibrating the axial physical coordinates of the cable under test based on the optical imaging data stream are as follows: using edge detection and feature recognition algorithms to process continuous image frames in the optical imaging data stream to identify and extract the outline of the cable under test in the image; based on the outline of the cable under test, determining the axis passing through the geometric center of the outline and defining the axis as the axial physical coordinate; and establishing a transformation relationship that maps the coordinates of each pixel in the thermal imaging data stream to the axial physical coordinate.

3. The method for testing the performance of high-temperature resistant cables according to claim 1, characterized in that, The steps for detecting concentration inflection points include: analyzing the concentration values ​​of the target component in the gas composition data stream. The process is performed to calculate the concentration value. First time derivative With the second time derivative Specifically, when the first-order time derivative and the second-order time derivative simultaneously satisfy the following conditions within the second detection window: and When the concentration inflection point is determined to have occurred, then... is a given positive threshold number used to characterize the effective growth rate of concentration.

4. The method for testing the performance of high-temperature resistant cables according to claim 1, characterized in that, The method further includes: synchronously acquiring acoustic emission data streams, which characterize high-frequency stress waves emitted by the cable due to internal microscopic fractures under thermal stress; and incorporating acoustic bursting events identified from the acoustic emission data streams as internal mechanical damage events into the pairing and formation process of the causal event chain, so that the final pyrolysis failure fingerprint contains event information that distinguishes between external morphological changes and internal structural damage.

5. The method for testing the performance of a high-temperature resistant cable according to claim 1, characterized in that, The method further includes: storing multiple generated pyrolysis failure fingerprints into a failure mode database, and using the failure mode database to train a time-series prediction model.

6. The method for testing the performance of high-temperature resistant cables according to claim 1, characterized in that, The duration of the second detection window is set based on the physical distance between the surface of the cable under test and the gas composition data stream acquisition point, as well as the physical delay time determined by the gas transmission rate of the target component under the test environment; the synchronous acquisition frequency is not less than 10 Hz; and the given tolerances are: the synchronization time jitter error is less than one sampling period, and the registration spatial positioning error is less than one pixel size.

7. The method for testing the performance of high-temperature resistant cables according to claim 1, characterized in that, The pyrolysis failure fingerprint is a structured data object. Within this structured data object, each event constituting a causal event chain records a unique event type identifier, an absolute timestamp of the event occurrence, the axial physical coordinates of the event occurrence, and the corresponding local temperature value and target component concentration value at the time of the event occurrence.

8. The method for testing the performance of high-temperature resistant cables according to claim 1, characterized in that, The specific steps for detecting crack events from the optical imaging data stream based on pixel gradient change rules are as follows: within the first detection window, the pixel gray-level gradient distribution along the radial direction of the cable under test in the optical imaging data stream is calculated; when a local peak appears in the pixel gray-level gradient distribution, and the value of the local peak exceeds the given crack judgment gradient threshold, the crack event is determined to have occurred at that location.

9. The method for testing the performance of a high-temperature resistant cable according to claim 1, characterized in that, All steps of the method are performed under the constraints of standard procedures and physical conditions that allow for uninterrupted continuous combustion testing.

10. A high-temperature resistant cable performance testing system, characterized in that, The system includes: A sensing and acquisition module, comprising a thermal imaging sensor, an optical camera, and a gas analyzer, is configured to synchronously acquire thermal imaging data streams, optical imaging data streams, and gas composition data streams at the same master clock during the continuous combustion test of a cable sample. A data processing module is connected to the sensing and acquisition module. The data processing module is configured to: calibrate the axial physical coordinates of the cable under test based on the optical imaging data stream, and register the thermal imaging data stream to these axial physical coordinates to establish a shared spatiotemporal domain; within this shared spatiotemporal domain, when the temperature at a certain location along the axial physical coordinate first reaches the glass transition temperature of the cable material... At that time, the first detection window and the second detection window are opened respectively, based on that time. Crack events are detected from the optical imaging data stream only within the first detection window according to the pixel gradient change rule, and the concentration inflection point of the target component is detected from the gas composition data stream only within the second detection window. Before performing time-series pairing, the synchronization and registration errors of the thermal imaging data stream, the optical imaging data stream, and the gas composition data stream are checked, and data frames with errors exceeding the given tolerance are discarded. Only the temperature rise, crack events, and concentration inflection points that pass the error closed-loop check are time-series paired to form a causal event chain. Based on the formed causal event chain, a pyrolysis failure fingerprint characterizing the dynamic failure characteristics of the cable is generated and output.

Images