Nuclear power high temperature seal assembly damage identification system

Abstract

The present application provides a nuclear power high-temperature sealing assembly damage identification system, and relates to the technical field of nondestructive testing.The present application solves the problem of difficult compromise between sensor tolerance and signal fidelity under high temperature of nuclear power through multi-dimensional heterogeneous perception and temperature change noise adaptive separation module: hardware thermal isolation is realized by using radiation-resistant optical fiber sensor and acoustic emission probe with waveguide rod, time-temperature synchronous marking and differential stripping strategy are combined, high-temperature noise is accurately removed, original high-frequency details are reserved, and noise floor is continuously monitored to judge sensor failure, false positives and false negatives are avoided, and equipment life is prolonged;further, a double-track feature identification module is designed: on the one hand, forward matching is carried out based on temperature stress matching template to generate second-level early warning to prevent weak features from being filtered out;on the other hand, energy transition and temperature change are directly monitored to generate first-level interruption instruction, and emergency shutdown and highest-level alarm are triggered through OR gate logic to realize millisecond-level response, guarantee early identification and predictive maintenance.

Description

Technical Field

[0001] This invention relates to the field of non-destructive testing technology, specifically a damage identification system for high-temperature sealing components in nuclear power plants. Background Technology

[0002] Nuclear power generation involves high temperatures, high pressures, and strong radiation, making sealing components a critical barrier against the leakage of radioactive media. Early sealing monitoring relied heavily on manual overhaul inspections or passive leak alarms based on temperature and humidity changes, which suffered from significant delays. With the development of non-destructive testing and artificial intelligence, the industry has gradually introduced sensing technologies such as acoustic emission and vision, striving to develop damage identification systems capable of capturing weak defect signals in real time under extreme operating conditions, with the aim of achieving a leap towards predictive maintenance.

[0003] However, existing damage identification systems still have significant shortcomings in practical applications. The most common pain point lies in the difficulty of balancing environmental tolerance and signal fidelity in front-end sensing devices. Under the long-term effects of high temperatures of hundreds of degrees Celsius and strong radiation within the nuclear island, traditional acoustic emission probes or piezoelectric sensing chips are prone to thermal degradation and material aging. This results in the initial signals acquired being mixed with a large amount of severe high-temperature thermal noise and even waveform distortion, significantly shortening the equipment lifespan and reducing the reliability of the raw data.

[0004] Given the prevalent problems of signal distortion and high noise at the hardware level, existing technologies typically attempt to passively compensate by introducing highly complex filtering and denoising models in the software. This leads to a subtle but highly detrimental secondary defect: "early weak feature obliteration and response delay caused by algorithm overcompensation." Because the system struggles to accurately separate extreme thermal noise and the weak high-frequency features of the initial initiation of microcracks in seals under complex operating conditions, powerful denoising algorithms often filter out the true early damage signals, resulting in "missed detections." Simultaneously, the complex compensation calculations significantly increase the edge processing load, causing the system to lose its millisecond-level real-time early warning capability when facing sudden damage expansion, severely violating the system's original intention of early identification. Summary of the Invention

[0005] The purpose of this invention is to provide a damage identification system for high-temperature sealing components in nuclear power plants, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A nuclear power plant high-temperature sealing component damage identification system includes:

[0008] The multidimensional heterogeneous sensing module is used to synchronously collect acoustic emission signals and real-time temperature gradient status of the sealed area under extreme operating conditions of the nuclear island by using a special sensing array closely attached to the sealing component. Each acoustic emission signal is assigned an independent timestamp and temperature stamp to establish a strong physical binding relationship between temperature and acoustic signals.

[0009] The temperature-variable noise adaptive separation module is used to receive acoustic emission signals with temperature stamps. Instead of the traditional global frequency filtering noise reduction method, it calls the pre-established "temperature thermal noise morphology topology library" based on the real-time temperature stamp and the service life of the sensor. It uses a differential strategy of logical subtraction to remove thermal noise and waveform distortion signals that match the current temperature in order to obtain a high-fidelity identification signal that retains the original high-frequency details.

[0010] The dual-track feature recognition module is used to perform concurrent logical processing on the signal to be identified with high fidelity, including establishing a transient mutation direct link and a weak feature fidelity matching link. The transient mutation direct link is used to avoid the edge computing load of complex algorithms and provide a delay-free response to sudden damage signals. The weak feature fidelity matching link is used to perform morphological comparison on the signal after stripping thermal noise, accurately identify the weak high-frequency features in the early stage of microcrack initiation, and avoid early signal annihilation caused by algorithm compensation.

[0011] The feedforward early warning and control module is used to execute a multi-level concurrent control scheme based on the logic state output by the dual-track feature identification module and the current operating condition command of the nuclear power unit, so as to realize early identification and early warning of the deterioration state of the sealing components and millisecond-level cut-off control of sudden leaks.

[0012] Furthermore, the multidimensional heterogeneous sensing module includes a hardware thermal isolation unit and a time-temperature synchronization marking unit;

[0013] The hardware thermal isolation unit consists of a radiation-resistant fiber optic temperature sensor and an acoustic emission probe with a flexible acoustic wave conduction waveguide rod. The waveguide rod physically isolates the core piezoelectric crystal of the acoustic emission probe from the extreme high-temperature region, reducing the initial thermal decay at the hardware level.

[0014] The time-temperature synchronization marking unit is used to lock and align the real-time temperature change curve acquired by the fiber optic temperature sensor with the original acoustic waveform acquired by the acoustic emission probe on the same time axis, generating an original composite signal bundle with dual markings of (time axis, temperature coordinates).

[0015] Furthermore, the temperature-varying noise adaptive separation module includes a benchmark library dynamic update unit and a differential stripping unit;

[0016] The benchmark library dynamic update unit is used to record the pure thermal noise waveform characteristics of sensors in the same batch under different temperature gradients and different service durations, and to construct a dynamically updated "temperature thermal noise morphology topology library" to map the thermal noise evolution law caused by sensor material aging.

[0017] The differential stripping unit is used to read the temperature coordinates in the original composite signal beam, retrieve the thermal noise baseline waveform that best matches the current temperature and service duration from the topology library, and use time-domain inverse superposition logic to cancel the original composite signal and the thermal noise baseline waveform, directly outputting the unfiltered, high-fidelity signal to be identified.

[0018] Furthermore, the dual-track feature recognition module includes a transient through-pass subunit, a morphology matching subunit, and a temperature stress mapping subunit;

[0019] The transient pass-through subunit constitutes the transient mutation pass-through link;

[0020] The morphological matching subunit and the temperature stress mapping subunit constitute the weak feature fidelity matching link;

[0021] Both links start simultaneously and operate independently upon receiving the fidelity identification signal to ensure robust system response under extreme conditions.

[0022] Furthermore, the transient pass-through subunit is pre-set with an original energy threshold and a temperature change slope, and the envelope energy transition rate of the fidelity signal to be identified and the synchronous temperature change slope are directly monitored.

[0023] When it is determined that the envelope energy transition rate exceeds the original energy threshold and the temperature change slope increases sharply, it is determined that the sealing component has suddenly ruptured or a large amount of media has leaked, and the first-level interrupt control command is directly generated to achieve millisecond-level response.

[0024] Furthermore, the temperature stress mapping subunit is used to lock the thermal expansion physical state of nuclear power sealing components, including flange gaskets and O-rings, at the current specific high temperature based on the current real-time temperature stamp, and accordingly retrieve the typical "early microcrack release frequency band" and "friction and wear waveform template" under the thermal expansion physical state; by limiting the identification range through physical state mapping, the retrieval load of the system in edge computing in massive data is adjusted.

[0025] Furthermore, the morphological matching subunit adopts forward matching logic; using the extracted high-fidelity signal to be identified, it directly performs multi-dimensional morphological overlap comparison with the retrieved "early microcrack release frequency band" or "friction and wear waveform template", including comparison of the logic sequence of rise time, duration and ring count;

[0026] If the overlap meets the preset feature logic lock, then it is determined that there is early minor damage, and a second-level warning instruction is generated.

[0027] Furthermore, the dual-track feature recognition module also includes a sensor health status logic judgment unit;

[0028] The sensor health status logic judgment unit is used to continuously track the noise floor amplitude of the thermal noise signal stripped from the temperature-varying noise adaptive separation module; when it is determined that the current noise floor amplitude is permanently increasing in a stepwise manner under the same temperature,

[0029] Furthermore, when the noise level exceeds the preset logic multiple of the sensor's initial calibration noise level, it is determined that the piezoelectric chip of the acoustic emission sensor itself has undergone irreversible thermal decay or material aging, generating a sensor failure replacement command to block false damage omissions and false alarms caused by sensor malfunctions.

[0030] Furthermore, the feedforward early warning and control module includes a logic decision-making unit and an instruction execution unit;

[0031] The logic decision unit uses "OR" gate logic to receive the first-level interrupt control command and uses "AND" gate logic to combine the unit status to receive the second-level early warning command;

[0032] The instruction execution unit is used to transform the decision results into specific system actions or alarm mechanisms without waiting for lengthy comprehensive numerical evaluation calculations.

[0033] Furthermore, the execution strategy of the logical decision-making unit is as follows:

[0034] If the first-level interrupt control command is received, the system skips all subsequent analysis, directly triggers the emergency shut-off linkage control of the relevant isolation valves in the primary or secondary loop of the nuclear island, and activates the highest-level central control room audible and visual alarm to resolve the response delay problem.

[0035] If only the second-level warning instruction is received and the first-level interrupt is not triggered, the system will mark the sealed node with a "yellow deterioration highlight" on the digital twin monitoring screen without stopping the stacking process, and automatically generate a predictive maintenance work order to record the damage evolution trend.

[0036] If a sensor failure replacement instruction is received, the system will temporarily transfer the monitoring weight of that node to a nearby redundant sensor and prompt maintenance personnel to replace the sensor probe first during the next major overhaul.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] This invention fundamentally solves the problem of balancing environmental tolerance and signal fidelity in front-end sensing devices under extreme high-temperature conditions in nuclear islands by employing a multi-dimensional heterogeneous sensing module and a temperature-variable noise adaptive separation module. The multi-dimensional heterogeneous sensing module utilizes a radiation-resistant fiber optic temperature sensor and an acoustic emission probe with a flexible acoustic waveguide rod to physically isolate the piezoelectric wafer from the high-temperature region, significantly reducing the risk of hardware thermal degradation and material aging. Simultaneously, it combines a time-temperature synchronization marking unit to generate an original composite signal bundle with dual markings of time axis and temperature coordinates. The temperature-variable noise adaptive separation module abandons global frequency filtering and utilizes a temperature thermal noise morphology topology library constructed by a benchmark library dynamic update unit. A differential stripping unit matches the thermal noise baseline waveform between the current temperature and service duration, employing a time-domain inverse superposition differential strategy to accurately strip away high-temperature thermal noise and distorted signals, obtaining a high-fidelity signal that retains original high-frequency details. Furthermore, the sensor health logic judgment unit continuously tracks the amplitude of the stripped thermal noise floor. When it determines that the amplitude is permanently increasing in a stepwise manner and exceeds the preset logic multiple of the initial calibrated noise floor, it generates a sensor failure replacement command to avoid false damage reports and missed detections. This linkage mechanism not only overcomes the problems of initial signal distortion and high noise, but also significantly extends the service life of special sensing equipment and ensures the reliability of the original data.

[0039] This invention also designs a dual-track feature recognition module to process the signals to be identified concurrently and independently. On one hand, the temperature stress mapping subunit locks the thermal expansion state of the sealing component based on the real-time temperature, retrieves the microcrack release frequency band and friction wear waveform template, and the morphology matching subunit uses forward matching logic to compare the signal with the template in terms of rise time, duration, ring count, etc., to generate a second-level early warning command. This avoids strong noise reduction filtering out the weak high-frequency features in the early stage of microcrack initiation in the sealing component, preventing missed detections. On the other hand, the transient pass-through subunit directly monitors the signal envelope energy transition rate and temperature change slope. When the energy exceeds the original energy threshold and shows a steep increase, it directly generates a first-level interrupt control command. Finally, the logic decision unit of the feedforward early warning and control module uses "OR" gate logic. After receiving the first-level interrupt command, it skips all subsequent analysis and the command execution unit triggers the emergency cut-off control of the isolation valves of the primary and secondary loops of the nuclear island and activates the highest-level central control room audible and visual alarm. This completely avoids complex edge computing and lengthy evaluation, achieving millisecond-level real-time response and fully implementing the original intention of early identification and predictive maintenance. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the overall method flow of the present invention. Detailed Implementation

[0041] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0042] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0043] Please see Figure 1 This invention provides a technical solution: a damage identification system for high-temperature sealing components in nuclear power plants, comprising:

[0044] The multidimensional heterogeneous sensing module is used to synchronously collect acoustic emission signals and real-time temperature gradient status of the sealed area under extreme operating conditions of the nuclear island by using a special sensing array closely attached to the sealing component. Each acoustic emission signal is assigned an independent timestamp and temperature stamp to establish a strong physical binding relationship between temperature and acoustic signals.

[0045] The temperature-variable noise adaptive separation module is used to receive acoustic emission signals with temperature stamps. Instead of the traditional global frequency filtering noise reduction method, it calls the pre-established "temperature thermal noise morphology topology library" based on the real-time temperature stamp and the service life of the sensor. It uses a differential strategy of logical subtraction to remove thermal noise and waveform distortion signals that match the current temperature in order to obtain a high-fidelity identification signal that retains the original high-frequency details.

[0046] The dual-track feature recognition module is used to perform concurrent logical processing on the signal to be identified with high fidelity, including establishing a transient mutation direct link and a weak feature fidelity matching link. The transient mutation direct link is used to avoid the edge computing load of complex algorithms and provide a delay-free response to sudden damage signals. The weak feature fidelity matching link is used to perform morphological comparison on the signal after stripping thermal noise, accurately identify the weak high-frequency features in the early stage of microcrack initiation, and avoid early signal annihilation caused by algorithm compensation.

[0047] The feedforward early warning and control module is used to execute a multi-level concurrent control scheme based on the logic state output by the dual-track feature identification module and the current operating condition command of the nuclear power unit, so as to realize early identification and early warning of the deterioration state of the sealing components and millisecond-level cut-off control of sudden leaks.

[0048] The multidimensional heterogeneous sensing module includes a hardware thermal isolation unit and a time-temperature synchronization marking unit;

[0049] The hardware thermal isolation unit consists of a radiation-resistant fiber optic temperature sensor and an acoustic emission probe with a flexible acoustic wave conduction waveguide rod. The waveguide rod physically isolates the core piezoelectric crystal of the acoustic emission probe from the extreme high-temperature region, reducing the initial thermal decay at the hardware level.

[0050] The time-temperature synchronization marking unit is used to lock and align the real-time temperature change curve acquired by the fiber optic temperature sensor with the original acoustic waveform acquired by the acoustic emission probe on the same time axis, generating an original composite signal bundle with dual markings of (time axis, temperature coordinates).

[0051] The temperature-varying noise adaptive separation module includes a benchmark library dynamic update unit and a differential stripping unit;

[0052] The benchmark library dynamic update unit is used to record the pure thermal noise waveform characteristics of sensors in the same batch under different temperature gradients and different service durations, and to construct a dynamically updated "temperature thermal noise morphology topology library" to map the thermal noise evolution law caused by sensor material aging.

[0053] The differential stripping unit is used to read the temperature coordinates in the original composite signal beam, retrieve the thermal noise baseline waveform that best matches the current temperature and service duration from the topology library, and use time-domain inverse superposition logic to cancel the original composite signal and the thermal noise baseline waveform, directly outputting the unfiltered, high-fidelity signal to be identified.

[0054] The dual-track feature recognition module includes a transient through-pass subunit, a morphology matching subunit, and a temperature stress mapping subunit;

[0055] The transient pass-through subunit constitutes the transient mutation pass-through link;

[0056] The morphological matching subunit and the temperature stress mapping subunit constitute the weak feature fidelity matching link;

[0057] Both links start simultaneously and operate independently upon receiving the fidelity identification signal to ensure robust system response under extreme conditions.

[0058] The transient pass-through subunit is pre-set with an original energy threshold and a temperature change slope, and directly monitors the envelope energy transition rate of the fidelity signal to be identified and the synchronous temperature change slope.

[0059] When it is determined that the envelope energy transition rate exceeds the original energy threshold and the temperature change slope increases sharply, it is determined that the sealing component has suddenly ruptured or a large amount of media has leaked, and the first-level interrupt control command is directly generated to achieve millisecond-level response.

[0060] The temperature stress mapping subunit is used to lock the thermal expansion physical state of nuclear power sealing components, including flange gaskets and O-rings, at the current specific high temperature based on the current real-time temperature stamp, and retrieve the typical "early microcrack release frequency band" and "friction and wear waveform template" under the thermal expansion physical state; by limiting the identification range through physical state mapping, the retrieval load of the system in edge computing in massive data is adjusted.

[0061] The morphological matching subunit adopts forward matching logic; using the extracted high-fidelity signal to be identified, it directly compares the multi-dimensional morphological overlap with the retrieved "early microcrack release frequency band" or "friction and wear waveform template", including the comparison of the logic sequence of rise time, duration and ring count;

[0062] If the overlap meets the preset feature logic lock, then it is determined that there is early minor damage, and a second-level warning instruction is generated.

[0063] The dual-track feature recognition module also includes a sensor health status logic judgment unit;

[0064] The sensor health status logic judgment unit is used to continuously track the noise floor amplitude of the thermal noise signal stripped from the temperature-varying noise adaptive separation module; when it is determined that the current noise floor amplitude is permanently increasing in a stepwise manner under the same temperature,

[0065] Furthermore, when the noise level exceeds the preset logic multiple of the sensor's initial calibration noise level, it is determined that the piezoelectric chip of the acoustic emission sensor itself has undergone irreversible thermal decay or material aging, generating a sensor failure replacement command to block false damage omissions and false alarms caused by sensor malfunctions.

[0066] The feedforward early warning and control module includes a logic decision-making unit and an instruction execution unit;

[0067] The logic decision unit uses "OR" gate logic to receive the first-level interrupt control command and uses "AND" gate logic to combine the unit status to receive the second-level early warning command;

[0068] The instruction execution unit is used to transform the decision results into specific system actions or alarm mechanisms without waiting for lengthy comprehensive numerical evaluation calculations.

[0069] The execution strategy of the logical decision-making unit is as follows:

[0070] If the first-level interrupt control command is received, the system skips all subsequent analysis, directly triggers the emergency shut-off linkage control of the relevant isolation valves in the primary or secondary loop of the nuclear island, and activates the highest-level central control room audible and visual alarm to resolve the response delay problem.

[0071] If only the second-level warning instruction is received and the first-level interrupt is not triggered, the system will mark the sealed node with a "yellow deterioration highlight" on the digital twin monitoring screen without stopping the stacking process, and automatically generate a predictive maintenance work order to record the damage evolution trend.

[0072] If a sensor failure replacement instruction is received, the system will temporarily transfer the monitoring weight of that node to a nearby redundant sensor and prompt maintenance personnel to replace the sensor probe first during the next major overhaul.

[0073] Furthermore, the core parameters used to drive the dual-track feature recognition and linkage control system include: the real-time temperature gradient entropy parameter (denoted as Entemp), the acoustic emission envelope energy transition rate parameter (denoted as Rae), the sensor aging degradation factor parameter (denoted as Dag), and the final output thermoacoustic coupling anomaly index parameter (denoted as Iano). Among these, the real-time temperature gradient entropy parameter Entemp is used to quantify the degree of disorder and irregularity of temperature fluctuations in a local area of ​​the sealing assembly within a short time window, in order to distinguish between normal unit temperature rise and abnormal thermal shock caused by high-pressure medium leakage. Its specific acquisition method relies on high-frequency temperature data collected by a radiation-resistant fiber optic temperature sensor closely attached to the sealing assembly, and its linguistic calculation model is adapted from the Shannon entropy model in information theory. Specifically:

[0074] Extract all continuous temperature sampling points within a fixed-length sliding time window preceding the current moment; calculate the temperature difference between adjacent sampling points; divide the range of all temperature differences within the time window into several equally wide intervals; count the frequency of temperature differences falling into each interval, and divide this frequency by the total number of sampling points within the window to obtain the probability value of each interval; for each probability value, calculate the product of the probability value and its natural logarithm; sum the product results corresponding to all probability values, and take the absolute value of the sum; divide this absolute value by the maximum theoretical logarithmic constant historically calibrated for the target seal under extreme rupture conditions, thereby quantizing and limiting the output value range to the 0-1 interval;

[0075] In this embodiment, 1000 temperature points are acquired within a sliding window with a time period of 100 milliseconds. After differential calculation and interval probability statistics, if the temperature fluctuations are extremely violent and disordered, the absolute value of the summation is close to the calibrated maximum logarithmic constant of 2.99. After division normalization, the output value of this parameter is 0.95. If it is in a normal shutdown state with a gradual increase, the output value is close to 0.05. This provides a normalization benchmark for eliminating accidental drift under extreme operating conditions.

[0076] The acoustic emission envelope energy transition rate parameter Rae characterizes the instantaneous surge in the envelope energy of the fidelity signal to be identified after differential stripping, and is a core dynamic indicator triggering transient abrupt changes in the direct link. Its determination relies on an energy integral ratio model combined with a logistic regression model for nonlinear compression; the linguistic calculation model is as follows:

[0077] Obtain the absolute values ​​of all amplitudes of the high-fidelity signal to be identified within the current microsecond-level time window; calculate the area integral of all amplitudes within the time window to obtain the current instantaneous energy value; obtain the arithmetic mean of the area integrals of the high-fidelity signal to be identified within the past ten steady-state time windows as the baseline energy value; subtract the baseline energy value from the current instantaneous energy value to obtain the energy difference, and divide the energy difference by the baseline energy value to obtain the initial transition ratio;

[0078] A logical activation process is constructed to map the value range to the interval between zero and one. This involves taking the natural constant as a power of the negative of the initial transition ratio to obtain an intermediate value. Then, the intermediate value is added to the constant to obtain the denominator variable. Finally, the constant is divided by the denominator variable to obtain the final acoustic emission envelope energy transition rate parameter Rae output value.

[0079] In this embodiment, when a sudden minor rupture occurs in the seal, the current instantaneous energy value surges to five times the baseline energy value. The initial transition ratio is four. After logic activation processing, its output value is smoothly compressed to 0.982, accurately reflecting the energy transition of extremely high risk and strictly satisfying dimensional convergence.

[0080] The sensor aging degradation factor parameter Dag is used to quantify the degree of irreversible performance degradation of the piezoelectric sensing chip under long-term high temperature and strong radiation, in order to block false damage signals and trigger an automatic transfer mechanism. Its calculation is based on the exponential decay law in the theory of material fatigue accumulation, as follows:

[0081] During each steady-state operation of the unit, the arithmetic mean of the current pure thermal noise base plate amplitude is extracted through the temperature-varying noise adaptive separation module; the pure thermal noise base plate amplitude of the same batch of sensors under the initial factory calibration state is extracted; the current pure thermal noise base plate amplitude is subtracted from the initial calibration pure thermal noise base plate amplitude to obtain the irreversible noise floor increment; the irreversible noise floor increment is divided by the system's preset maximum tolerance noise floor drift threshold to obtain the linear decay ratio; boundary convergence processing is constructed, that is, if the linear decay ratio is greater than a constant 1, it is directly taken as a constant 1; if it is less than or equal to a constant 1, the linear decay ratio is the final output value of the factor. Furthermore, the preset maximum tolerance noise floor drift threshold is 20 millivolts. If the current noise floor is 10 millivolts higher than the factory noise floor, the output value is 0.5, representing that the sensor has entered the middle of its lifespan.

[0082] Based on the aforementioned key parameters, the calculation process in this embodiment is decomposed into a rigorous sequential execution logic. First, basic data and status input are executed. The system receives a raw composite signal bundle with dual tags generated by the hardware thermal isolation unit. Simultaneously, the sensor aging degradation factor parameter Dag, calculated in the previous cycle, and the unit's current operating mode identifier code are initialized in the digital twin control kernel. Next, in the dynamic differential and fidelity signal extraction stage, the system reads the real-time temperature stamp from the raw composite signal bundle, retrieves and extracts the thermal noise baseline waveform that best matches the temperature and service duration from the topology library of the benchmark library dynamic update unit. Then, the time-domain inverse superposition logic is executed, subtracting the amplitude value of each time sampling point of the raw acoustic waveform from the amplitude value of the corresponding thermal noise baseline waveform. This directly outputs a fidelity signal that fully preserves high-frequency transient details, and simultaneously extracts the amplitude of this baseline waveform to update the current sensor aging degradation factor parameter Dag.

[0083] After completing the signal fidelity extraction, the system initiates dual-track concurrent computation and state interception. In the first concurrent branch, the transient pass-through subunit, the system does not perform any morphological extraction on the signal. Instead, it directly performs calculations on the fidelity-identified signal and the synchronous fiber temperature signal, outputting the aforementioned acoustic emission envelope energy transition rate parameter Rae and the real-time temperature gradient entropy parameter Entemp, and passing these two parameters to the fusion evaluator. Simultaneously, in the second concurrent branch, the weak feature fidelity matching subunit, the system calls the temperature stress mapping subunit, locks the waveform template library based on temperature stamps, and performs temporal domain overlap matching between the envelope rise time and ring count of the fidelity-identified signal and the extracted template. If the matching sequence exceeds the set logic lock length, the "early weak degradation flag" is set to true in memory.

[0084] To address the challenge of false triggering of the single-dimensional threshold under extreme nuclear island conditions due to local thermofluid disturbances, the system introduces a "dynamic nonlinear weighted Euclidean distance fusion method" in the fusion evaluator to calculate the thermoacoustic coupling anomaly index parameter Iano. This mechanism breaks through the limitations of traditional linear weighting. The specific computational process is as follows: The input acoustic emission envelope energy transition rate parameter Rae and the real-time temperature gradient entropy parameter Entemp are obtained respectively; the acoustic emission envelope energy transition rate parameter Rae is subtracted from a constant to obtain the first deviation, and its square is calculated to obtain the first deviation squared value; the first deviation squared value is multiplied by a preset static weighting constant 0.6 (this constant is calibrated based on the leading characteristics of acoustic wave energy released in high-pressure, high-temperature destructive simulation) to obtain the acoustic distance component; the real-time temperature gradient entropy parameter Entemp is subtracted from a constant and... The second deviation squared value is obtained by squaring the second deviation squared value and multiplying it by the preset thermal hysteresis weight constant 0.4 to obtain the thermal distance component. The acoustic distance component and the thermal distance component are added together to obtain the Euclidean distance square sum, and the square root is taken to obtain the comprehensive distance scalar. The comprehensive distance scalar is subtracted from the constant one to obtain the final output thermoacoustic coupling anomaly index parameter Iano. This logic ensures that the anomaly index will exhibit an accelerated nonlinear response approaching the constant one if and only if the temperature disorder and the acoustic destructive force simultaneously approach the extreme level, thus forming a highly robust judgment criterion.

[0085] Based on the multi-dimensional fusion calculation results, the system enters the collaborative control and action output stage. According to hardware-level discrimination logic: if the sensor aging degradation factor parameter Dag is determined to be equal to a constant, the system triggers a sensor failure and redundancy handover command; if the thermoacoustic coupling anomaly index parameter Iano is greater than the set critical mutation constant, the system will ignore all morphological matching results and directly generate a first-level interrupt control command; if the anomaly index parameter is less than the critical mutation constant, and the "early weak degradation flag" in the second concurrent branch is true, a second-level warning command is generated. The feedforward control module then converts the command into physical actions, either by directly routing the first-level interrupt control command to the servo control interface of the nuclear island isolation valve for millisecond-level emergency cutoff, or by encoding and sending the second-level warning command to the digital twin monitoring screen for degradation highlighting and work order distribution, truly achieving multi-level concurrent collaborative gain control.

[0086] It should be noted that all calculation formulas in this application employ regression analysis, including but not limited to machine learning algorithms, to deeply analyze the collected parameters and identify their natural trends and interrelationships. Specialized software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Then, cross-validation and other methods are used to objectively evaluate the model performance, and continuous feedback and optimization are combined to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their effectiveness and accuracy. In all calculation formulas in this application, the parameters in each formula undergo dimensionless processing within a consistent range to ensure that different physical quantities are compared on the same scale; dimensionless processing techniques include, but are not limited to, min-max-normalization and Z-score standardization.

[0087] The technical solution of this invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random-access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments of this invention.

[0088] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0089] 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A damage identification system for high-temperature sealing components in nuclear power plants, characterized in that, include: The multidimensional heterogeneous sensing module is used to synchronously collect acoustic emission signals and real-time temperature gradient status of the sealed area under extreme operating conditions of the nuclear island by using a special sensing array closely attached to the sealing component. Each acoustic emission signal is assigned an independent timestamp and temperature stamp to establish a strong physical binding relationship between temperature and acoustic signals. The temperature-variable noise adaptive separation module is used to receive acoustic emission signals with temperature stamps. Instead of the traditional global frequency filtering noise reduction method, it calls the pre-established "temperature thermal noise morphology topology library" based on the real-time temperature stamp and the service life of the sensor. It uses a differential strategy of logical subtraction to remove thermal noise and waveform distortion signals that match the current temperature in order to obtain a high-fidelity identification signal that retains the original high-frequency details. The dual-track feature recognition module is used to perform concurrent logical processing on the signal to be identified with high fidelity, including establishing a transient mutation direct link and a weak feature fidelity matching link; the transient mutation direct link is used to avoid the edge computing load of complex algorithms and to provide a delay-free response to sudden damage signals; The weak feature fidelity matching link is used to perform morphological comparison on the signal after stripping thermal noise, accurately identifying weak high-frequency features in the early stage of microcrack initiation, and avoiding early signal obliteration caused by algorithm compensation. Specifically, the dual-track feature recognition module includes a transient pass-through subunit, a morphological matching subunit, and a temperature stress mapping subunit. The transient pass-through subunit constitutes the transient mutation pass-through link. The morphological matching subunit and the temperature stress mapping subunit constitute the weak feature fidelity matching link. The two links start simultaneously and run independently after receiving the fidelity identification signal to ensure the robustness of the system response under extreme conditions. The transient pass-through subunit is preset with an original energy threshold and a temperature change slope. It directly monitors the envelope energy transition rate of the fidelity signal to be identified and the synchronous temperature change slope. When it is determined that the envelope energy transition rate exceeds the original energy threshold and the temperature change slope increases sharply, it is determined that the sealing component has suddenly ruptured or a large amount of media has leaked. The first-level interrupt control command is directly generated to achieve millisecond-level response. The temperature stress mapping subunit is used to lock the thermal expansion physical state of nuclear power sealing components, including flange gaskets and O-rings, at the current specific high temperature based on the current real-time temperature stamp, and retrieve the typical "early microcrack release frequency band" and "friction and wear waveform template" under the thermal expansion physical state; by limiting the identification range through physical state mapping, the retrieval load of the system in edge computing in massive data is adjusted. The morphological matching subunit adopts forward matching logic; using the extracted high-fidelity signal to be identified, it directly compares the multi-dimensional morphological overlap with the retrieved "early microcrack release frequency band" or "friction and wear waveform template", including the comparison of the logic sequence of rise time, duration and ring count; if the overlap meets the preset feature logic lock, it is determined that there is early weak damage and a second-level warning instruction is generated. The feedforward early warning and control module is used to execute a multi-level concurrent control scheme based on the logic state output by the dual-track feature identification module and the current operating condition command of the nuclear power unit, so as to realize early identification and early warning of the deterioration state of the sealing components and millisecond-level cut-off control of sudden leaks.

2. The nuclear power high-temperature sealing assembly damage identification system according to claim 1, characterized in that: The multidimensional heterogeneous sensing module includes a hardware thermal isolation unit and a time-temperature synchronization marking unit; The hardware thermal isolation unit consists of a radiation-resistant fiber optic temperature sensor and an acoustic emission probe with a flexible acoustic wave conduction waveguide rod. The waveguide rod physically isolates the core piezoelectric crystal of the acoustic emission probe from the extreme high-temperature region, reducing the initial thermal decay at the hardware level. The time-temperature synchronization marking unit is used to lock and align the real-time temperature change curve acquired by the fiber optic temperature sensor with the original acoustic waveform acquired by the acoustic emission probe on the same time axis, generating an original composite signal bundle with dual markings of (time axis, temperature coordinates).

3. The nuclear power high-temperature sealing assembly damage identification system according to claim 2, characterized in that: The temperature-varying noise adaptive separation module includes a benchmark library dynamic update unit and a differential stripping unit; The benchmark library dynamic update unit is used to record the pure thermal noise waveform characteristics of sensors in the same batch under different temperature gradients and different service durations, and to construct a dynamically updated "temperature thermal noise morphology topology library" to map the thermal noise evolution law caused by sensor material aging. The differential stripping unit is used to read the temperature coordinates in the original composite signal beam, retrieve the thermal noise baseline waveform that best matches the current temperature and service duration from the topology library, and use time-domain inverse superposition logic to cancel the original composite signal and the thermal noise baseline waveform, directly outputting the unfiltered, high-fidelity signal to be identified.

4. The nuclear power high-temperature sealing assembly damage identification system according to claim 3, characterized in that: The dual-track feature recognition module also includes a sensor health status logic judgment unit; The sensor health status logic judgment unit is used to continuously track the noise floor amplitude of the thermal noise signal stripped from the temperature-varying noise adaptive separation module; when it is determined that the current noise floor amplitude is permanently increasing in a stepwise manner under the same temperature, Furthermore, when the noise level exceeds the preset logic multiple of the sensor's initial calibration noise level, it is determined that the piezoelectric chip of the acoustic emission sensor itself has undergone irreversible thermal decay or material aging, generating a sensor failure replacement command to block false damage omissions and false alarms caused by sensor malfunctions.

5. The nuclear power high-temperature sealing assembly damage identification system according to claim 4, characterized in that: The feedforward early warning and control module includes a logic decision-making unit and an instruction execution unit; The logic decision unit uses "OR" gate logic to receive the first-level interrupt control command and uses "AND" gate logic to combine the unit status to receive the second-level early warning command; The instruction execution unit is used to transform the decision results into specific system actions or alarm mechanisms without waiting for lengthy comprehensive numerical evaluation calculations.

6. The nuclear power high-temperature sealing assembly damage identification system according to claim 5, characterized in that: The execution strategy of the logical decision-making unit is as follows: If the first-level interrupt control command is received, the system skips all subsequent analysis, directly triggers the emergency shut-off linkage control of the relevant isolation valves in the primary or secondary loop of the nuclear island, and activates the highest-level central control room audible and visual alarm to resolve the response delay problem. If only the second-level warning instruction is received and the first-level interrupt is not triggered, the system will mark the sealed node with a "yellow deterioration highlight" on the digital twin monitoring screen without stopping the stacking process, and automatically generate a predictive maintenance work order to record the damage evolution trend. If a sensor failure replacement instruction is received, the system will temporarily transfer the monitoring weight of that node to a nearby redundant sensor and prompt maintenance personnel to replace the sensor probe first during the next major overhaul.

Images