A biaxial creep deformation monitoring system for nuclear fuel cladding tubes

Abstract

The present invention relates to the field of deformation monitoring technology, and specifically to a biaxial creep deformation monitoring system for nuclear fuel cladding tubes. The system comprises: an acquisition and monitoring module, a deformation identification module, a sample aggregation module, a deformation training module, and a risk warning module. The present invention screens out deformation mutation sections based on multiple parameters, and further extracts the same-direction change trends of various physical parameters within the same interval to construct synergistic change segments, thereby accurately identifying the physical causes of structural mutations in cladding tubes. On this basis, the time difference between the response inflection point time and the deformation rate peak time of different parameters is quantitatively analyzed to form a difference pattern reflecting the system response lag, thereby enhancing the temporal dynamic perception of deformation behavior. By extracting samples with different lag characteristics, reconstructing the linkage relationship point set between parameters, and combining Euclidean distance to construct a multidimensional response interval, clustering and classification of deformation density and change trend are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of deformation monitoring, and in particular to a nuclear fuel cladding tube biaxial creep deformation monitoring system. Background Art

[0002] The field of deformation monitoring technology encompasses technologies used to monitor and control physical or environmental parameters in real time, particularly in industrial and engineering applications. Core aspects of this area include monitoring and feedback control of physical quantities such as cladding tube axial temperature, pressure, displacement, deformation, and vibration. Monitoring technology collects sensor data, combines data processing and analysis, and tracks and adjusts the system's operating status in real time, ensuring safe, stable, and optimized equipment operation. Monitoring systems typically include data acquisition equipment, signal processing units, and data storage and display interfaces. They are widely used in aerospace, energy, manufacturing, and other fields for equipment monitoring and fault warning.

[0003] The nuclear fuel cladding tube biaxial creep deformation monitoring system is designed to monitor the biaxial creep deformation of nuclear fuel cladding tubes under high-temperature and high-pressure conditions. This system utilizes specific sensors to monitor the biaxial deformation of nuclear fuel cladding tubes in real time during long-term use in nuclear reactors. Using precise measurement methods, the system acquires real-time deformation data during reactor operation, ensuring the reliability and accuracy of the monitoring data. This system continuously tracks the operational status of nuclear fuel cladding tubes within nuclear reactors, providing data support for operational management and enabling timely action when anomalies occur.

[0004] Traditional systems rely heavily on basic sensors to collect single-parameter time-series data for trend identification in cladding tube deformation monitoring, ignoring the importance of coordinated changes between parameters and response time differences in analyzing deformation behavior. Under actual high-temperature and high-pressure conditions, cladding tube deformation is often caused by the combined effects of multiple physical fields. Existing technologies lack multi-angle identification methods for deformation precursors, making it impossible to conduct in-depth analysis and trend judgment of the deformation process. For example, monitoring only through deformation rate curves can lead to misjudgments or delayed responses when local stress changes have not yet been significantly reflected in surface deformation. In addition, existing systems rely on fixed threshold settings for anomaly identification, making it difficult to adapt to the parameter fluctuations in nuclear reactor operation, resulting in frequent omissions or false alarms in complex coupling situations. The lack of an identification mechanism for coupling trends between parameters also limits the ability to model risk evolution paths, making it impossible to effectively support traceability analysis and precise early warning of abnormal trends, affecting the safety and reliability of overall operational management. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a nuclear fuel cladding tube biaxial creep deformation monitoring system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a nuclear fuel cladding tube biaxial creep deformation monitoring system, the system comprising:

[0007] The acquisition and monitoring module obtains four types of time series data generated in real time under biaxial working conditions: the outer surface deformation rate of the nuclear fuel cladding tube, the axial temperature of the cladding tube, the gas pressure in the cladding tube cavity, and the stress on the inner wall of the cladding tube. It selects the deformation mutation sections and extracts the multi-parameter unidirectional change intervals to construct a multi-parameter coordinated time series segment group.

[0008] The deformation recognition module calculates the time difference between the inflection point time of the cladding tube axial temperature, the cladding tube inner wall stress, the cladding tube inner cavity gas pressure and the peak time of the cladding tube outer surface deformation rate based on the multi-parameter coordinated time sequence segment group, and generates a response hysteresis difference pattern group;

[0009] The sample aggregation module performs cluster label division according to the hysteresis type of each hysteresis structure in the response hysteresis difference pattern group and reconstructs the four-dimensional feature distribution to generate a parameter delay structure feature set;

[0010] The deformation training module calls each type of lag label sample in the parameter delay structure feature set, reconstructs the joint value point set of the parameters in the deformation response time period, marks the section with the most intensive fluctuation as the main deformation response band, and generates a deformation density response interval set.

[0011] The improvements of the present invention are that the multi-parameter coordinated time series segment group includes a deformation rate change section of the outer surface of the cladding tube, a temperature fluctuation trend section of the axial direction of the cladding tube, a stress growth trend section of the inner wall of the cladding tube, and a stable section of the gas pressure disturbance in the inner cavity of the cladding tube; the response lag difference mode group is specifically the cladding tube multi-parameter response time difference distribution type, the cladding tube parameter maximum response offset structure, and the cladding tube local response lag mean square error value; the parameter delay structure feature set includes the deformation dominant response sample type, the four-parameter integral offset distribution structure, and the four-parameter feature interval fluctuation coefficient; the deformation density response interval set is specifically the cladding tube multi-parameter response clustering sample structure, the parameter fluctuation density interval within the response time period, and the main deformation response segment position index.

[0012] The present invention is improved in that the acquisition and monitoring module includes:

[0013] The rate mutation identification submodule collects the time series data of the outer surface deformation rate of the nuclear fuel cladding tube under biaxial loading conditions, calculates the rate of change of the deformation rate at consecutive time points, extracts the first-order derivative value curve of the change rate, screens the extreme points of the derivative curve, identifies the time segment where the deformation rate change rate has a sudden change, marks the start and end indexes of the target time segment, and generates a set of deformation rate mutation time intervals;

[0014] The parameter fluctuation extraction submodule extracts the cladding tube axial temperature time series data, cladding tube inner wall stress data, and cladding tube inner cavity gas pressure data within the corresponding time period based on the deformation rate mutation time interval set, calculates the temperature fluctuation gradient per unit time, the slope change per unit time of the inner wall stress, and the fluctuation continuity index of the inner cavity pressure within the target time period, aligns the three types of parameter fluctuation data according to the time index and performs distribution annotation to generate a multi-parameter synchronous fluctuation structure group;

[0015] The collaborative segment marking submodule calls the multi-parameter synchronous fluctuation structure group, compares the change direction of each parameter in the same time period with the concentrated deformation rate change direction of the deformation rate mutation time interval, determines whether the directions of the three parameters are consistent with the deformation rate change direction, selects all time periods that meet the direction consistency, integrates the start and end times of the target time period and the time series segments of the four parameters, and generates a multi-parameter collaborative time series segment group.

[0016] The present invention is improved in that the deformation recognition module includes:

[0017] The time series construction submodule extracts the original sampling data of the outer surface deformation rate of the inner cladding tube, the axial temperature of the cladding tube, the inner wall stress of the cladding tube, and the gas pressure in the inner cavity of the cladding tube for each section based on the multi-parameter coordinated time series segment group, uniformly adjusts the parameter sampling frequency to a consistent interval length, reorders each parameter according to a unified time step, and generates a standard sampling time series group;

[0018] The response time difference calculation submodule calls the standard sampling time series group, locates the inflection point of the cladding tube axial temperature, cladding tube inner wall stress, and cladding tube inner cavity gas pressure in each segment of data, records the time difference between the corresponding inflection point time and the peak time of the deformation rate of the cladding tube outer surface in the same segment, calculates the mean square error of the three types of parameter time difference values ​​respectively and integrates them into a structure set to generate a multi-parameter response time difference feature group;

[0019] The response structure statistics submodule extracts the distribution of the response time differences of the three types of parameters in each segment of data based on the multi-parameter response time difference feature group, identifies the parameter combination corresponding to the maximum response time difference, counts the location of the maximum response difference in each segment, extracts the corresponding parameter sequence, integrates the maximum response difference combination, time position index and difference amplitude, and generates a response lag difference pattern group.

[0020] The present invention has the following improvements: the unsampled data is filled by linear interpolation method, using the formula:

[0021] ;

[0022] Calculate the standard sampling point time to be interpolated respectively Parameters on , reorder each interpolated parameter according to a uniform time step to construct a standardized time series data set and generate a standard sampling time series group;

[0023] in, is the standard sampling point time to be interpolated, For the data, Parameters of the two known sampling time points on the left and right; Close to The known sampling time points on the left and right sides, Represents a function of a parameter at a specified point in time.

[0024] The present invention is improved in that the sample aggregation module includes:

[0025] The hysteresis segment extraction submodule extracts the start and end time of the corresponding segment in the cladding tube original data according to the hysteresis structure identifier recorded in the response hysteresis difference pattern group, establishes a classification label for each segment of data according to the hysteresis type, and summarizes the original sequences of the cladding tube outer surface deformation rate, cladding tube axial temperature, cladding tube inner wall stress and cladding tube inner cavity gas pressure in the segment according to the label to generate a hysteresis structure label data set;

[0026] The integral difference calculation submodule calls the hysteresis structure label data set, performs difference calculation on the gradient function of the deformation rate of the outer surface of the cladding tube in each type of sample and the cladding tube axial temperature change curve, the cladding tube inner wall stress growth curve, and the cladding tube inner cavity gas pressure disturbance curve, respectively, performs statistics on the integral value of the difference function on the unified time axis, and extracts the mean and standard fluctuation rate of the integral offset interval in each type of sample to generate an integral offset group;

[0027] The characteristic configuration reconstruction submodule extracts the dominant offset sequence of the deformation rate of the outer surface of the cladding tube and the other three types of parameter change structures according to the integral mean and volatility distribution of each category of samples in the integral offset group to construct a unified index, reorganizes them into a four-dimensional space feature point set according to the hysteresis type, constructs a feature interval sequence for each category, integrates them into a structured numerical output, and generates a parameter delay structure feature set.

[0028] The present invention is improved in that the deformation training module includes:

[0029] The sample screening submodule calls all samples with hysteresis labels in the parameter delay structure feature set, extracts the integral offset value of the deformation rate of the outer surface of the cladding tube and the stress of the inner wall of the cladding tube, judges the offset direction, classifies samples with positive correlation in the offset direction into the same group, marks and extracts the cladding tube axial temperature and inner cavity gas pressure parameter segments of the samples, and generates a positive offset sample set;

[0030] The response clustering submodule extracts the joint value of four parameters, namely, the deformation rate of the outer surface of the cladding tube, the stress of the inner wall of the cladding tube, the axial temperature of the cladding tube, and the gas pressure in the inner cavity of the cladding tube, in each sample segment based on the forward offset sample set, constructs a four-dimensional parameter value point set within the deformation response time period, calculates the Euclidean distance between two specified points in the point set, divides the cluster groups and records the value boundaries of each group to generate a multi-parameter response cluster interval group;

[0031] The density extraction submodule counts the time density of the four parameter change points in each category according to the clustering structure in the multi-parameter response clustering interval group, extracts the density distribution sequence and calculates the local maximum position, marks the time period where the maximum value is located as the area with the most intensive deformation change, integrates and numbers the target time segments, and generates a deformation density response interval set.

[0032] The present invention has the following improvements: for calculating the Euclidean distance between two specified points in the point set, the formula is used:

[0033] ;

[0034] in, Indicates that in the sample segment In, With the The Euclidean distance between the sampling points, The material is Sampling point time The deformation rate gradient after normalization, The material is Sampling point time The normalized stress on the inner wall is The material is Sampling point time The normalized temperature distribution in the axial direction is The material is Sampling point time The normalized pressure in the cavity, The material is Time points of adjacent sampling points The deformation rate gradient after normalization, The material is Time points of adjacent sampling points The normalized stress on the inner wall is The material is Time points of adjacent sampling points The normalized temperature distribution in the axial direction is The material is Time points of adjacent sampling points The normalized pressure in the cavity.

[0035] The present invention is improved in that it further includes a risk warning module, which tracks, based on each main response band in the deformation density response interval concentration, whether a downward trend curve of the gas pressure in the cladding tube cavity in a subsequent adjacent time window forms a positive coupling with an increasing slope of the deformation rate of the cladding tube outer surface, and determines an abnormal section to generate a deformation coupling abnormal section set;

[0036] The deformation coupling abnormal section set includes the forward coupling period of the cladding tube inner cavity gas pressure and deformation rate, the linkage over-limit interval of the cladding tube axial temperature and inner wall stress, and the frequency segment number and time distribution density of multi-parameter superposition abnormalities.

[0037] The present invention is improved in that the risk warning module includes:

[0038] The coupling relationship identification submodule obtains each main response band in the deformation density response interval concentration, extracts the cladding tube inner cavity gas pressure decrease trend curve and the cladding tube outer surface deformation rate increase slope sequence in the adjacent time window after the end of the main response band, compares the slope directions of the two point by point, selects the time segment that constitutes forward coupling, and generates a forward coupling time series group;

[0039] The limit superposition judgment submodule calls each time segment in the forward coupling time series group, obtains the axial temperature change gradient of the cladding tube and the stress growth amplitude of the inner wall of the cladding tube in the corresponding period, numerically superimposes the two types of parameter values, compares the superposition result with the set multi-parameter coupling threshold, filters the time index segment that exceeds the superposition limit, and generates an over-limit superposition segment set;

[0040] The abnormal segment statistical extraction submodule marks the time index of the over-limit superposition segment set, counts the occurrence frequency and cumulative concentration of each segment within the continuous monitoring period, integrates and outputs the segments whose occurrence frequency is higher than the set frequency threshold and whose concentration density exceeds the limited interval, establishes a mapping between the corresponding time index and the parameter sequence number, and generates a deformation coupling abnormal segment set.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are:

[0042] In the present invention, by screening out deformation mutation sections based on multi-parameter variables, and further extracting the same-direction change trends of various physical parameters in the same interval, and constructing synergistic change segments, the physical inducement of the cladding tube structure mutation can be accurately locked. On this basis, the time difference between the response inflection point time and the deformation rate peak time of different parameters is quantitatively analyzed to form a difference pattern reflecting the system response lag, thereby enhancing the time dynamic perception ability of deformation behavior. By extracting samples with different hysteresis characteristics, reconstructing the linkage relationship point set between parameters, and combining Euclidean distance to construct a multi-dimensional response interval, the deformation density and change trend are clustered and classified, and the main response period of deformation evolution is accurately delineated. This processing logic based on multi-parameter time series configuration analysis and response clustering effectively improves the analytical ability of the multi-factor coupled deformation process of the cladding tube. At the same time, with the help of the subsequent parameter fluctuation trend in the deformation density interval to identify the positive coupling relationship, combined with the superposition strength judgment threshold, the intelligent identification and risk warning of the multi-parameter mutation coupling area are successfully realized. The overall solution enhances the recognition accuracy and early perception capability of abnormal sections through the dual characterization of structural change trends and parameter coupling paths, breaking through the limitation of traditional monitoring methods that rely solely on single-parameter abnormal alarms, and constructing a multi-dimensional, high-resolution structural deformation monitoring framework. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a system module diagram proposed by the present invention;

[0044] Figure 2 This is the system framework diagram proposed by the present invention;

[0045] Figure 3 This is a schematic diagram of the data acquisition and monitoring module of the present invention;

[0046] Figure 4 is a schematic diagram of a deformation recognition module of the present invention;

[0047] Figure 5 is a schematic diagram of a sample aggregation module of the present invention;

[0048] Figure 6 is a schematic diagram of a deformation training module of the present invention;

[0049] Figure 7 Schematic diagram of the risk warning module of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0051] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operate in a specific direction, and therefore should not be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0052] See also Figure 1 The present invention provides a technical solution: a nuclear fuel cladding tube biaxial creep deformation monitoring system, the system comprising:

[0053] The acquisition and monitoring module obtains four types of time series data generated in real time under biaxial working conditions: the outer surface deformation rate of the nuclear fuel cladding tube, the axial temperature of the cladding tube, the gas pressure in the cladding tube cavity, and the stress on the inner wall of the cladding tube. It selects the deformation mutation sections and extracts the multi-parameter unidirectional change intervals to construct a multi-parameter coordinated time series segment group.

[0054] The deformation recognition module calculates the time difference between the inflection point time of the cladding tube axial temperature, the inner wall stress of the cladding tube, the gas pressure in the cladding tube cavity and the peak time of the deformation rate of the outer surface of the cladding tube based on the multi-parameter coordinated time sequence segment group, and generates a response hysteresis difference pattern group;

[0055] The sample aggregation module divides the cluster labels according to the lag type of each lag structure in the response lag difference pattern group and reconstructs the four-dimensional feature distribution to generate a parameter delay structure feature set;

[0056] The deformation training module calls each type of lag label sample in the parameter delay structure feature set, reconstructs the joint value point set of the parameters in the deformation response time period, marks the section with the most intensive fluctuation as the main deformation response band, and generates a deformation density response interval set;

[0057] The multi-parameter collaborative time series segment group includes the deformation rate change section of the outer surface of the cladding tube, the axial temperature fluctuation trend section of the cladding tube, the stress growth trend section of the inner wall of the cladding tube, and the stable section of the gas pressure disturbance in the inner cavity of the cladding tube; the response lag difference mode group specifically includes the multi-parameter response time difference distribution type of the cladding tube, the maximum response offset structure of the cladding tube parameters, and the lag mean square error value of the local response of the cladding tube; the parameter delay structure feature set includes the deformation dominant response sample type, the four-parameter integral offset distribution structure, and the four-parameter characteristic interval fluctuation coefficient; the deformation density response interval set specifically includes the multi-parameter response clustering sample structure of the cladding tube, the parameter fluctuation density interval within the response time period, and the main deformation response segment position index.

[0058] See also Figure 2 and Figure 3 , the acquisition and monitoring module includes:

[0059] The rate mutation identification submodule collects the time series data of the outer surface deformation rate of the nuclear fuel cladding tube under biaxial loading conditions, calculates the rate of change of the deformation rate at consecutive time points, extracts the first-order derivative value curve of the change rate, screens the extreme points of the derivative curve, identifies the time segment where the deformation rate change rate has a sudden change, marks the start and end indexes of the target time segment, and generates a set of deformation rate mutation time intervals;

[0060] The deformation rate data of the cladding tube is collected. This process needs to be carried out in a biaxial loading test environment. The experimental equipment applies Axis and The composite force in the axial direction, a high-frequency laser vibrometer is installed on the surface of the cladding tube. The deformation displacement is recorded in seconds and converted into deformation rate ,in Indicates time (Unit: seconds) The deformation rate of the outer surface of the cladding tube at the time (unit: mm / s), Represents the time interval between two consecutive samplings (unit: seconds). First, calculate the rate of change of the deformation rate at adjacent time points. The formula is: ,in Indicates the rate of change of deformation rate at time t (unit: mm / s²), Indicates time The deformation rate value is further derived from the rate of change to obtain: ,in represents the derivative of the rate of change over time t (i.e., the rate of change of the rate of change, in mm / s³), Indicates time The rate of change of the extreme point is determined by sliding the window (e.g., the size is 5 points) to determine whether the extreme point meets the conditions. (maximum) or (minimum value), the selected extreme value points are further used for mutation determination and setting the mutation recognition threshold , which is calculated as: ,in Represents a derivative sequence The mean value (unit: mm / s³), Represents the standard deviation of the derivative sequence. If a certain extreme point Exceed It is identified as a mutation point, such as mm / s³, mm / s³, the threshold is mm / s³, records the start and end time indexes of all mutation segments that meet the conditions =1.23 seconds to = 1.67 seconds, and all mutation time periods are sorted to form a deformation rate mutation time interval set for subsequent processing.

[0061] The parameter fluctuation extraction submodule extracts the cladding tube axial temperature time series data, cladding tube inner wall stress data, and cladding tube inner cavity gas pressure data within the corresponding time period based on the deformation rate mutation time interval set. It calculates the temperature fluctuation gradient per unit time, the slope change per unit time of the inner wall stress, and the fluctuation continuity index of the inner cavity pressure within the target time period. The three types of parameter fluctuation data are aligned by time index and distributed annotated to generate a multi-parameter synchronous fluctuation structure group.

[0062] After obtaining the deformation mutation time interval set, the axial temperature of the cladding tube in the corresponding time period is extracted , inner wall stress and intracavitary pressure Three types of parameter fluctuation data, among which represents the axial temperature of the cladding tube at time t (unit: Celsius), measured by a distributed infrared sensor array; The stress on the inner wall of the cladding tube at time t (unit: MPa) is obtained by converting the deformation collected by the strain gauge using Hooke's law; The gas pressure in the cladding tube cavity at time t (unit: kPa) is obtained by a high-frequency response micro pressure sensor. The three types of parameter fluctuation data are processed according to the following formula: The temperature fluctuation gradient per unit time is ,in represents the temperature fluctuation rate at time t, is the time interval; the stress slope per unit time is ,in is the stress slope change; the cavity pressure fluctuation continuity is: ,in is the average fluctuation degree of cavity pressure in the mutation section (unit: kPa), Indicates the The cavity pressure value at each sampling point, It represents the absolute difference in pressure between two adjacent moments. For the summation sign, all pressure changes starting from the second point are accumulated. is the total number of sampling points in the mutation time period. For example, if the time period is 1.23s–1.67s and the sampling frequency is 100Hz, we get , all fluctuation data are aligned according to the same time index to generate multiple sets of ternary structures , representing the three types of parameter states at each time point, which are further used for structure annotation and collaborative analysis.

[0063] The collaborative segment marking submodule calls the multi-parameter synchronous fluctuation structure group, compares the change direction of each parameter in the same time period with the change direction of the deformation rate in the deformation rate mutation time interval, determines whether the direction of the three parameters is consistent with the deformation rate change direction, selects all time periods that meet the direction consistency, integrates the start and end times of the target time period and the time series segments of the four parameters, and generates a multi-parameter collaborative time series segment group;

[0064] Read the multi-parameter synchronous fluctuation structure group and calculate the temperature fluctuation gradient for each time index t , stress change rate , cavity pressure fluctuation value and deformation rate The consistency of the change direction is judged by using the symbol function Direction identification, where :like Then it returns +1 (indicating an increase), Returns -1 (indicates decrease). Then return (Indicates stability), the direction is considered consistent if the following conditions are met: ,in They are temperature gradient, stress change rate, cavity pressure fluctuation value and deformation rate, all of which are data at time t. Represents the data value of the previous moment. If the change direction of the four items at two adjacent sampling points is completely consistent, it is determined to be a coordinated segment. Continuous segments with the same judgment value are extracted, such as [1.25s, 1.45s] and [1.50s, 1.62s]. 21 groups of sampling data (sampling frequency 100Hz) are extracted from each segment to form a four-dimensional data structure: ,These structures are numbered and summarized into multi-parameter coordinated timing fragment groups for further processing by downstream modules.

[0065] See also Figure 2 and Figure 4 , the deformation recognition module includes:

[0066] The time series construction submodule extracts the original sampling data of the outer surface deformation rate of the inner cladding tube, the axial temperature of the cladding tube, the inner wall stress of the cladding tube, and the gas pressure in the cladding tube cavity of each section based on the multi-parameter coordinated time series segment group, uniformly adjusts the parameter sampling frequency to a consistent interval length, reorders each parameter according to a unified time step, and generates a standard sampling time series group;

[0067] Based on the generated multi-parameter coordinated time series segment group, directly extract the four types of parameter data already available in each segment ,in :Indicates the time (unit: seconds) the outer surface deformation rate of the cladding tube at the time (unit: mm / s), : is the axial temperature change gradient per unit time (in degrees Celsius / second), : is the slope of the inner wall stress growth (in MPa / s), : is the continuous fluctuation of the gas pressure in the cavity (dimensionless). The task of this module is to unify the sampling frequency, standardize the time step and reorder the time series in all data segments. For example, set the unified sampling frequency to , then the time step is , if the start and end times of a data segment are , , then the length of this segment is , the corresponding number of standard sampling points is calculated as: ,in: : The number of standard time sampling points that should be in this section; : Indicates rounding down operation; : unified time step, in seconds; : Respectively represent the start and end time of the original data segment (in seconds, the English subscript is used to distinguish it from the variable t); then, construct the standard time index sequence as follows: ,in: : No. The standard sampling time of each time point (in seconds); : Sampling point sequence subscript (positive integer, from 1 to 201); reorder all parameter data points according to this sequence. If the original data is in some If the upper part is not sampled, linear interpolation is required to fill it in. The interpolation formula is: , the meaning of the formula is explained as follows: :Target time to be sought A parameter on Any one of ); : Standard sampling point time to be interpolated (unit: seconds); :In the original data, sandwiched between Parameter values ​​at two known sampling time points on the left and right; : respectively close to The known sampling time points on the left and right sides (unit: seconds); : represents any parameter (such as ) at a certain point in time; for example, in the original data , , now need to calculate Substituting the value of into the formula, we get: , so the interpolated value of this parameter at 2.01 seconds is 3.3. Repeat this process to complete the data filling at all standard time points in this section. After performing this operation on all four parameter sequences, the time step is formed. , a standard sampling time series group with consistent length and aligned index, which is used for cross-parameter time response comparison and difference extraction processing in subsequent modules.

[0068] The response time difference calculation submodule calls the standard sampling time series group to locate the inflection point of the cladding tube axial temperature, cladding tube inner wall stress, and cladding tube inner cavity gas pressure in each segment of data, and records the time difference between the corresponding inflection point time and the peak time of the deformation rate of the cladding tube outer surface in the same segment. The mean square error of the three types of parameter time difference values ​​is calculated and integrated into a structure set to generate a multi-parameter response time difference feature group;

[0069] Call the standard sampling time series group, without changing the original sequence structure, and analyze the three types of parameter sequences in each segment of data: cladding tube axial temperature, cladding tube inner wall stress, and cladding tube inner cavity gas pressure, and locate the inflection point of each type of parameter, that is, identify the temperature sequence , stress series , cavity pressure sequence The position where the value of the median changes dramatically, and the corresponding time point is recorded as the temperature inflection point time , stress inflection point time , air pressure inflection point time ,in 、 、 Respectively represent the response inflection point appearance time of temperature, stress and pressure changes, t represents the standardized sampling time in seconds; then extract the deformation rate sequence of the outer surface of the cladding tube from the same section The peak time point in ,in represents the deformation rate value at time t (in mm / s), For this paragraph The time point at which the maximum value is reached. Then the response time difference between the inflection point time of the three types of parameter changes and the rate peak time is calculated and expressed as: 、 、 ,in 、 、 It represents the response time difference of temperature, stress and pressure relative to the peak value of rate, in seconds, and the symbol represents the absolute value function; in all This operation is performed on each time series segment to obtain the response time difference of each type of parameter in each segment. ,in , Indicates the The segment data number is a positive integer. Then, the mean square error of the response time difference of each type of parameter is calculated to measure the fluctuation degree of the response time. The calculation formula is: ,in: : parameter type x ( Indicates temperature, represents stress, represents the mean square error of the response time differences in all segments (in seconds²); : total number of sample segments; : No. The response time difference of parameter x in the segment, in seconds; : The mean time difference of the response of parameter x, calculated as ; : Indicates the sum operation of all segments; : Segment number index. Assume that the temperature response time difference in the three segments is , then: , Repeat this operation for and The two types of parameters are calculated separately 、 ,The three mean square error results are integrated to form a multi-parameter response time difference feature group, which provides input for the response structure statistics submodule.

[0070] The response structure statistics submodule extracts the distribution of the response time differences of the three types of parameters in each data segment based on the multi-parameter response time difference feature group, identifies the parameter combination corresponding to the maximum response time difference, counts the location of the maximum response difference in each segment, extracts the corresponding parameter sequence, integrates the maximum response difference combination, time position index and difference amplitude, and generates a response lag difference pattern group;

[0071] Based on the multi-parameter response time difference feature group, the response time difference of the three types of parameters in each segment of data is 、 、 Compare and select the one with the largest value to define it as the maximum response time difference, which represents the physical parameter with the longest response in this section. The formula is: ,in Indicates the maximum value operation among the three, and records the parameter type corresponding to the maximum value. , for example, if Maximum, then ; At the same time, mark the data segment number where the response difference is located as , indicating its position in the entire data, extracting the original parameter sequence of this segment for subsequent analysis and verification; organizing the maximum response difference combination into a structural entry ,like: The stress response difference of the 8th segment is the largest, with a value of 0.23 seconds. Next, to determine whether there is a significant response lag, the response difference threshold is defined. , calculated as: ,in is the average response time difference, is the standard deviation, here we can directly set the threshold of the temperature parameter to =0.12s, the threshold value of stress parameter is =0.15s, the threshold value of the air pressure parameter is =0.11s, if the maximum response difference of a certain section satisfies , then the segment is classified into the response hysteresis difference mode group, indicating that the physical response has significant hysteresis characteristics; for example, if , then the segment constitutes a hysteresis mode data item, and finally all the structure entries that meet the hysteresis conditions will be represented by the maximum response parameter type, the segment number The combination of , difference amplitude and response lag difference pattern group is used for abnormal behavior recognition, fault warning and model input feature set construction in downstream modules.

[0072] See also Figure 2 and Figure 5 , the sample aggregation module includes:

[0073] The hysteresis segment extraction submodule extracts the start and end times of the corresponding segment in the cladding tube raw data based on the hysteresis structure identifier recorded in the response hysteresis difference pattern group. It establishes classification labels for each segment of data according to the hysteresis type, and summarizes the original sequences of the cladding tube outer surface deformation rate, cladding tube axial temperature, cladding tube inner wall stress, and cladding tube inner cavity gas pressure in the segment according to the label to generate a hysteresis structure label dataset.

[0074] Read the data structure of each record in the response lag difference pattern group and extract the structure type fields in turn , paragraph index field , hysteresis value field ,in Indicates the hysteresis type to which the record belongs (such as temperature hysteresis , stress hysteresis , air pressure hysteresis ), For the The segment number corresponding to the sample in the data set is a positive integer. It is the maximum response time difference of the sample record in this section, in seconds. Enter the original data set and use the index positioning operation to find the time starting point of the corresponding number segment and the end point For example, in a record , get the start and end times through the timestamp index , Then, the time window slicing method is used to extract all the original sampling points in the interval, where each time point is bound to a parameter quadruple: the deformation rate of the outer surface of the cladding tube , axial temperature , inner wall stress , Inner cavity pressure , subscript Indicates the Standard sampling time points, with time as the primary key, and the original data points constitute vector pairs , all data by lag type Cluster storage, if a record type is , then the segment it belongs to is classified as "stress hysteresis". After the system completes the traversal of all segments, the entire data structure is divided into multiple sub-datasets according to the hysteresis label. Each sample maintains an independent segment number, an independent time axis and a full sequence of four parameters, and finally forms a hysteresis structure label dataset for subsequent integral analysis and processing.

[0075] The integral difference calculation submodule calls the hysteresis structure label dataset and performs difference calculation on the gradient function of the deformation rate of the outer surface of the cladding tube in each type of sample, the axial temperature change curve of the cladding tube, the stress growth curve of the inner wall of the cladding tube, and the gas pressure disturbance curve of the inner cavity of the cladding tube. The integral value of the difference function on the unified time axis is statistically analyzed, and the mean and standard fluctuation rate of the integral offset interval in each type of sample are extracted to generate an integral offset group.

[0076] Traverse all sample segments grouped by hysteresis type in the hysteresis structure label dataset, and extract the deformation rate of the outer surface of the cladding tube segment by segment for each type of sample The original sequence, where the subscript Indicates the standard time sampling point, The sampling time of the point is in seconds. Use the central difference method to calculate its gradient function , for example, The point sampling value is =1.2 mm / s, the next point is = 1.5 mm / s, time step , then the gradient of this point is approximately , and then read the other three original sequences in this segment from the same time axis: temperature gradient , stress growth , air pressure disturbance , and make a difference between it and the rate gradient function to construct the difference function ,in , each type of difference function In a unified timeline The integral is performed on the time interval , the integration is completed by numerical approximation, and the integral of each segment of data under the difference function is obtained. For example, the calculation is performed on a sample of a "temperature lag" segment, and the integral result is , indicating that the total deviation between temperature and rate gradient is 0.26 units. After all samples are processed in sequence, the integral value sequence is classified and counted according to the hysteresis type, which are recorded as , where the superscript Indicates the Segment sample, sub-subscript Indicates the parameter type. After classification, the sample mean and standard fluctuation rate of each type of offset integral value are calculated. For example, for the pressure lag sample, the integral mean is , the volatility is ,This type of statistical structure will be used for feature configuration operations in the next module.

[0077] The characteristic configuration reconstruction submodule extracts the dominant offset sequence of the deformation rate of the outer surface of the cladding tube and the other three types of parameter change structures based on the integral mean and volatility distribution of each category of samples in the integral offset group to construct a unified index. It is reorganized into a four-dimensional spatial feature point set according to the hysteresis type, and a characteristic interval sequence is constructed for each type. The sequence is integrated into a structured numerical output to generate a parameter delay structure feature set.

[0078] Read the integral mean corresponding to each type of lagged sample in the integral offset group With standard volatility , where the letters Represents three hysteresis type parameters: temperature, stress, and pressure. Each hysteresis type contains multiple samples. Each sample is a sequence set within a period of time. For each segment, the deformation rate gradient sequence Perform normalization and convert each item into Interval, after processing, each sample sequence corresponds to multiple standard time points , the other three parameters Extract the offset vector according to the corresponding type and perform offset processing: For example, for temperature hysteresis type samples, in each Construct eigenvectors on ,in is the mean of the integral offset of the temperature hysteresis class; the first dimension in this vector represents the unit normalized value of the dominant variable deformation rate, the second dimension is the temperature offset value, and the third and fourth dimensions are the original values ​​of the other two types of physical quantities. All feature points are arranged in chronological order to form a feature point sequence within the sample segment. The sample number is , the feature point number is , then the overall characteristic structure is the point set , each feature point is , the unit and source parameters of each dimension in the four-dimensional vector are clearly traceable. Finally, a unified index feature interval is constructed for samples of all lag types, and the categories to which they belong are marked respectively, so as to realize the unified four-dimensional spatial expression under different lag mechanisms and integrate them into a parameter delay structure feature set.

[0079] See also Figure 2 and Figure 6 ,The deformation training module includes:

[0080] The sample screening submodule calls all samples with hysteresis labels in the parameter delay structure feature set, extracts the integral offset value of the deformation rate of the cladding tube outer surface and the stress of the cladding tube inner wall, determines the offset direction, and groups samples with positive correlation in the offset direction into the same group. It then marks and extracts the cladding tube axial temperature and inner cavity gas pressure parameter segments of the samples to generate a positive offset sample set.

[0081] Call all samples in the parameter delay structure feature set that have been marked with lag labels, and the sample number is set to , extract the corresponding time series points in each segment Each sample contains the deformation rate gradient of the outer surface of the cladding tube , cladding tube inner wall stress , axial temperature and the inner cavity gas pressure A continuous sequence of , also including pre-calculated integral offset values ,in Indicates the In the sample and The overall offset between the sequences, in units of integral value, is used to reflect the correlation between the change directions of the two. When judging the offset direction, the judgment threshold is set to The value is set based on the fact that the integral offset is essentially the sum of the differences between the two parameters, and its positive or negative value directly reflects the direction relationship. When , it means that the deformation rate is generally higher than the stress response. Therefore, setting the critical point to zero can effectively distinguish positive correlation from negative correlation without standardization. In the example, if there are samples The integral offset is , then the sample is judged as a positive offset sample, and the module classifies it into the positive set. Then, the temperature series in the same time period of the sample is With air pressure series Perform a complete extraction, along with the segment number , sample start and end time, and four-parameter feature structure are packaged together to form a record in the forward migration sample set. This processing flow is for all After the sample is completed, the output is a structured sample subset containing positive labels, number indexes, and original temperature and pressure sequences.

[0082] Based on the forward offset sample set, the response clustering submodule extracts the joint value of four parameters in each sample segment: the deformation rate of the outer surface of the cladding tube, the stress of the inner wall of the cladding tube, the axial temperature of the cladding tube, and the gas pressure in the inner cavity of the cladding tube. It constructs a four-dimensional parameter value point set within the deformation response time period, calculates the Euclidean distance between two specified points in the point set, divides the cluster groups, and records the value boundaries of each group to generate a multi-parameter response cluster interval group.

[0083] Traverse the forward offset sample set numbered For each sample segment, extract each sampling point in its time series Four parameters on deformation rate gradient , inner wall stress , axial temperature , cavity air pressure , forming a four-dimensional joint point , and constitute the response point set , and then perform Euclidean distance calculation on all adjacent pairs of points in the sample point set, using the formula: ,in Indicates that in the sample segment In, With the The Euclidean distance between sampling points, all dimension units are normalized to dimensionless data to facilitate clustering processing, for example: if in the The point value is 、 、 、 , the next point is 、 、 、 , then substitute: , set the cluster radius threshold to This value refers to the median of the average distance distribution of 80% of the adjacent points in the normalized sample point set and then relaxes 20% upward to ensure that the clustering includes regular fluctuation samples while excluding isolated points. Finally, all the Euclidean distances less than The sample point set is classified into the same cluster, and the number, time interval and boundary values ​​of the four parameters of each cluster are output to form a multi-parameter response cluster interval group.

[0084] The density extraction submodule calculates the temporal density of the four parameter change points in each category based on the cluster structure in the multi-parameter response cluster interval group, extracts the density distribution sequence and calculates the location of the local maximum value, marks the time period where the maximum value is located as the area with the most intensive deformation change, integrates and numbers the target time segments, and generates a deformation density response interval set;

[0085] Read the cluster number of each type in the multi-parameter response cluster interval group in turn All sample segments and extract each time point in the segment The corresponding four parameters are: , density analysis is performed on the four parameters respectively, and density is defined as the unit time interval The statistical result of the number of internal parameter mutations is that the parameter value sequence is compared point by point in each sliding window. If the absolute value of the difference between adjacent sampling points exceeds the set mutation benchmark value, (Unit is normalized value), it is counted as a mutation. The threshold is set based on the mean of the standard deviation of all sample points. , and multiply by 2 to get the offset limit. Example: If the stress parameter sequence 、 ,but , recorded as one mutation, the density statistics method is: in each 0.1 second sliding window, record the number of mutations and get the density value , the unit is times / second, if a stress segment is within the window If 6 mutations are recorded in times / second, and then search for the local maximum point in the full density sequence. The judgment rule is: if and , then the time point is considered to be a local maximum, and the time period of the point is recorded and numbered. Finally, all maximum value segments are integrated into a deformation density response interval set. The output format is an interval structure group consisting of three items: number, time interval, and dominant mutation parameter type.

[0086] See also Figure 2 and Figure 7 , also includes a risk warning module, which concentrates each main response band according to the deformation density response interval, tracks whether the downward trend curve of the gas pressure in the cladding tube cavity in the subsequent adjacent time window is positively coupled with the increasing slope of the deformation rate of the cladding tube outer surface, and judges the abnormal section to generate a deformation coupling abnormal section set;

[0087] The deformation coupling abnormal section set includes the forward coupling period of the cladding tube inner cavity gas pressure and deformation rate, the linkage exceeding limit interval of the cladding tube axial temperature and inner wall stress, and the frequency segment number and time distribution density of multi-parameter superposition abnormality.

[0088] The risk warning module includes:

[0089] The coupling relationship identification submodule obtains each main response band in the deformation density response interval, extracts the downward trend curve of the gas pressure in the cladding tube cavity and the upward slope sequence of the deformation rate on the outer surface of the cladding tube in the adjacent time window after the end of the main response band, compares the slope directions of the two point by point, selects the time segments that constitute forward coupling, and generates a forward coupling time series group;

[0090] Get the deformation density response interval numbered as Each main response band of the segment, record its end time point in each segment , extend the adjacent time window interval backward ,in To fix the extension window length, this value comes from the average response offset time of the coupling lag section in the historical data, which is obtained through the statistics of the sample response curve to ensure that the complete post-response behavior is covered. The gas pressure sequence of the cladding tube cavity is extracted within this window. The deformation rate gradient sequence of the outer surface of the cladding tube ,in Indicates the first Standard sampling time points, in seconds, call the sliding window difference function to calculate the instantaneous slope of the pressure and rate gradients at each pair of adjacent sampling points. The calculation formula is: ,in is the parameter name, To unify the sampling step, in the example, if When the pressure drops from 2.8kPa to 2.6kPa and the velocity gradient increases from 1.1 to 1.4mm / s², we have , , the two directions are opposite, satisfying , which is the forward coupling decision point. The system compares point by point in the sample segment according to this method, filters out all the time index points that continuously meet the above conditions, and merges them into continuous forward coupling time segments to form the segment After processing, all samples are integrated into a forward coupling time series group.

[0091] The limit superposition judgment submodule calls each time segment in the forward coupling time series group to obtain the axial temperature gradient of the cladding tube and the stress growth amplitude of the inner wall of the cladding tube during the corresponding period, numerically superimposes the two types of parameter values, compares the superposition results with the set multi-parameter coupling threshold, and filters the time index segments that exceed the superposition limit to generate a set of over-limit superposition segments;

[0092] Each segment in the traversal forward coupled time series group is numbered time segments, corresponding to the time series Axial temperature gradient of the cladding tube The stress growth rate of the inner wall of the cladding tube Perform one-to-one extraction and perform sum operation at each time point to construct multi-parameter superposition response value ,in The unit is degrees Celsius / second + MPa / second, no unit conversion is required due to subsequent normalization processing, and then combined with the set coupling superposition threshold For comparison, the threshold is defined as: the sum of the normalized means of the temperature and stress parameters plus twice the joint standard deviation, that is, , where each statistic is obtained by analyzing the training sample set, and 、 、 、 , substituting into This value is the lower limit standard for determining whether it is a severe synchronous response. In the example, if at a certain time point 、 ,but , judged as overrun superposition points, all overrun points are clustered by time index to form continuous segments, and the output is a set of overrun superposition segments after merging, and the maximum value in each segment is retained. Value and start and end index positions.

[0093] The abnormal segment statistical extraction submodule marks the time index of the over-limit superposition segment set, counts the occurrence frequency and cumulative concentration of each segment within the continuous monitoring period, integrates and outputs the segments whose occurrence frequency exceeds the set frequency threshold and whose concentration density exceeds the limited interval, establishes a mapping between the corresponding time index and the parameter sequence number, and generates a deformation coupling abnormal segment set;

[0094] Read the paragraph numbers in the overrunning overlay fragment set in sequence and time index collection , within the designated monitoring period Count the frequency of occurrence of each segment within the period and cumulative time ratio ,in Indicates the number of times the segment is judged as an over-limit segment in all detection cycles. Indicates the proportion of its total occurrence time in the monitoring interval and sets the frequency threshold , concentration threshold , frequency threshold The upper quartile values ​​of the frequency distribution of each segment in the sample set are rounded up. For example, in the analysis, the average frequency of the segment is 1.8, and the 75th percentile is 2.5, which is rounded up to ,and The source is the mean of the distribution of the proportion of concentrated time in each period Adding standard deviation ,Right now , the system for each segment Make a judgment on and , it is marked as an abnormal segment, for example: a segment The frequency of occurrence is 5 times, accounting for the total time ,determined as an abnormal segment, the module integrates all the segment numbers, time indexes, and parameter structure indexes that meet the judgment conditions and outputs them as a set of deformation coupling abnormal segments.

[0095] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A nuclear fuel cladding tube biaxial creep deformation monitoring system, characterized in that: The system comprises: The acquisition and monitoring module obtains four types of time series data generated in real time under biaxial working conditions: the outer surface deformation rate of the nuclear fuel cladding tube, the axial temperature of the cladding tube, the gas pressure in the cladding tube cavity, and the stress on the inner wall of the cladding tube. It selects the deformation mutation sections and extracts the multi-parameter unidirectional change intervals to construct a multi-parameter coordinated time series segment group. The deformation recognition module calculates the time difference between the inflection point time of the cladding tube axial temperature, the cladding tube inner wall stress, the cladding tube inner cavity gas pressure and the peak time of the cladding tube outer surface deformation rate based on the multi-parameter coordinated time sequence segment group, and generates a response hysteresis difference pattern group; The sample aggregation module performs cluster label division according to the hysteresis type of each hysteresis structure in the response hysteresis difference pattern group and reconstructs the four-dimensional feature distribution to generate a parameter delay structure feature set; The deformation training module calls each type of lag label sample in the parameter delay structure feature set, reconstructs the joint value point set of the parameters in the deformation response time period, marks the section with the most intensive fluctuation as the main deformation response band, and generates a deformation density response interval set; The multi-parameter coordinated time series segment group includes the deformation rate change section of the outer surface of the cladding tube, the axial temperature fluctuation trend section of the cladding tube, the stress growth trend section of the inner wall of the cladding tube, and the stable section of the gas pressure disturbance in the inner cavity of the cladding tube; the response lag difference mode group specifically includes the cladding tube multi-parameter response time difference distribution type, the cladding tube parameter maximum response offset structure, and the cladding tube local response lag mean square error value; the parameter delay structure feature set includes the deformation dominant response sample type, the four-parameter integral offset distribution structure, and the four-parameter feature interval fluctuation coefficient; the deformation density response interval set specifically includes the cladding tube multi-parameter response clustering sample structure, the parameter fluctuation density interval within the response time period, and the main deformation response segment position index; The deformation recognition module includes: The time series construction submodule extracts the original sampling data of the outer surface deformation rate of the inner cladding tube, the axial temperature of the cladding tube, the inner wall stress of the cladding tube, and the gas pressure in the inner cavity of the cladding tube for each section based on the multi-parameter coordinated time series segment group, uniformly adjusts the parameter sampling frequency to a consistent interval length, fills the unsampled data through the linear interpolation method, reorders each parameter according to a unified time step, and generates a standard sampling time series group; The response time difference calculation submodule calls the standard sampling time series group, locates the inflection point of the cladding tube axial temperature, cladding tube inner wall stress, and cladding tube inner cavity gas pressure in each segment of data, records the time difference between the corresponding inflection point time and the peak time of the deformation rate of the cladding tube outer surface in the same segment, calculates the mean square error of the three types of parameter time difference values ​​respectively and integrates them into a structure set to generate a multi-parameter response time difference feature group; The response structure statistics submodule extracts the distribution of the response time differences of the three types of parameters in each segment of data based on the multi-parameter response time difference feature group, identifies the parameter combination corresponding to the maximum response time difference, counts the location of the maximum response difference in each segment, extracts the corresponding parameter sequence, integrates the maximum response difference combination, time position index and difference amplitude, and generates a response lag difference pattern group.

2. The nuclear fuel cladding tube biaxial creep deformation monitoring system according to claim 1, characterized in that: The acquisition and monitoring module includes: The rate mutation identification submodule collects the time series data of the outer surface deformation rate of the nuclear fuel cladding tube under biaxial loading conditions, calculates the rate of change of the deformation rate at consecutive time points, extracts the first-order derivative value curve of the change rate, screens the extreme points of the derivative curve, identifies the time segment where the deformation rate change rate has a sudden change, marks the start and end indexes of the target time segment, and generates a set of deformation rate mutation time intervals; The parameter fluctuation extraction submodule extracts the cladding tube axial temperature time series data, cladding tube inner wall stress data, and cladding tube inner cavity gas pressure data within the corresponding time period based on the deformation rate mutation time interval set, calculates the temperature fluctuation gradient per unit time, the slope change per unit time of the inner wall stress, and the fluctuation continuity index of the inner cavity pressure within the target time period, aligns the three types of parameter fluctuation data according to the time index and performs distribution annotation to generate a multi-parameter synchronous fluctuation structure group; The collaborative segment marking submodule calls the multi-parameter synchronous fluctuation structure group, compares the change direction of each parameter in the same time period with the concentrated deformation rate change direction of the deformation rate mutation time interval, determines whether the directions of the three parameters are consistent with the deformation rate change direction, selects all time periods that meet the direction consistency, integrates the start and end times of the target time period and the time series segments of the four parameters, and generates a multi-parameter collaborative time series segment group.

3. The nuclear fuel cladding tube biaxial creep deformation monitoring system according to claim 1, characterized in that: The unsampled data is filled in by linear interpolation method, using the formula: ； Calculate the standard sampling point time to be interpolated respectively Parameters on , the parameters include temperature fluctuation gradient, stress change rate, cavity pressure fluctuation value and deformation rate. Each parameter after interpolation is reordered according to a unified time step to construct a standardized time series data set and generate a standard sampling time series group; in, is the standard sampling point time to be interpolated, For the data, Parameters of the two known sampling time points on the left and right; Close to The known sampling time points on the left and right sides, Represents a function of a parameter at a specified point in time.

4. The nuclear fuel cladding tube biaxial creep deformation monitoring system according to claim 1, characterized in that: The sample aggregation module includes: The hysteresis segment extraction submodule extracts the start and end time of the corresponding segment in the cladding tube original data according to the hysteresis structure identifier recorded in the response hysteresis difference pattern group, establishes a classification label for each segment of data according to the hysteresis type, and summarizes the original sequences of the cladding tube outer surface deformation rate, cladding tube axial temperature, cladding tube inner wall stress and cladding tube inner cavity gas pressure in the segment according to the label to generate a hysteresis structure label data set; The integral difference calculation submodule calls the hysteresis structure label data set, performs difference calculation on the gradient function of the deformation rate of the outer surface of the cladding tube in each type of sample and the cladding tube axial temperature change curve, the cladding tube inner wall stress growth curve, and the cladding tube inner cavity gas pressure disturbance curve, respectively, performs statistics on the integral value of the difference function on the unified time axis, and extracts the mean and standard fluctuation rate of the integral offset interval in each type of sample to generate an integral offset group; The characteristic configuration reconstruction submodule extracts the dominant offset sequence of the deformation rate of the outer surface of the cladding tube and the other three types of parameter change structures according to the integral mean and volatility distribution of each category of samples in the integral offset group to construct a unified index, reorganizes them into a four-dimensional space feature point set according to the hysteresis type, constructs a feature interval sequence for each category, integrates them into a structured numerical output, and generates a parameter delay structure feature set.

5. The nuclear fuel cladding tube biaxial creep deformation monitoring system according to claim 4, characterized in that: The deformation training module includes: The sample screening submodule calls all samples with hysteresis labels in the parameter delay structure feature set, extracts the integral offset value of the deformation rate of the outer surface of the cladding tube and the stress of the inner wall of the cladding tube, judges the offset direction, classifies samples with positive correlation in the offset direction into the same group, marks and extracts the cladding tube axial temperature and inner cavity gas pressure parameter segments of the samples, and generates a positive offset sample set; The response clustering submodule extracts the joint value of four parameters, namely, the deformation rate of the outer surface of the cladding tube, the stress of the inner wall of the cladding tube, the axial temperature of the cladding tube, and the gas pressure in the inner cavity of the cladding tube, in each sample segment based on the forward offset sample set, constructs a four-dimensional parameter value point set within the deformation response time period, calculates the Euclidean distance between two specified points in the point set, divides the cluster groups and records the value boundaries of each group to generate a multi-parameter response cluster interval group; The density extraction submodule counts the time density of the four parameter change points in each category according to the clustering structure in the multi-parameter response clustering interval group, extracts the density distribution sequence and calculates the local maximum position, marks the time period where the maximum value is located as the area with the most intensive deformation change, integrates and numbers the target time segments, and generates a deformation density response interval set.

6. The nuclear fuel cladding tube biaxial creep deformation monitoring system according to claim 5, characterized in that: To calculate the Euclidean distance between two specified points in the point set, the formula is used: ； in, Indicates that in the sample segment In With the The Euclidean distance between the sampling points, The material is Sampling point time The deformation rate gradient after normalization, The material is Sampling point time The normalized stress on the inner wall is The material is Sampling point time The normalized temperature distribution in the axial direction is The material is Sampling point time The normalized pressure in the cavity, The material is Time points of adjacent sampling points The deformation rate gradient after normalization, The material is Time points of adjacent sampling points The normalized stress on the inner wall is The material is Time points of adjacent sampling points The normalized temperature distribution in the axial direction is The material is Time points of adjacent sampling points The normalized pressure in the cavity.

7. The nuclear fuel cladding tube biaxial creep deformation monitoring system according to claim 1, characterized in that: The invention also includes a risk warning module, which tracks whether the downward trend curve of the gas pressure in the inner cavity of the cladding tube in the subsequent adjacent time window is positively coupled with the increasing slope of the deformation rate of the outer surface of the cladding tube according to each main response band in the deformation density response interval concentration, and determines the abnormal section to generate a deformation coupling abnormal section set; The deformation coupling abnormal section set includes the forward coupling period of the cladding tube inner cavity gas pressure and deformation rate, the linkage over-limit interval of the cladding tube axial temperature and inner wall stress, and the frequency segment number and time distribution density of multi-parameter superposition abnormalities.

8. The nuclear fuel cladding tube biaxial creep deformation monitoring system according to claim 7, characterized in that: The risk warning module includes: The coupling relationship identification submodule obtains each main response band in the deformation density response interval concentration, extracts the cladding tube inner cavity gas pressure decrease trend curve and the cladding tube outer surface deformation rate increase slope sequence in the adjacent time window after the end of the main response band, compares the slope directions of the two point by point, selects the time segment that constitutes forward coupling, and generates a forward coupling time series group; The limit superposition judgment submodule calls each time segment in the forward coupling time series group, obtains the axial temperature change gradient of the cladding tube and the stress growth amplitude of the inner wall of the cladding tube in the corresponding period, numerically superimposes the two types of parameter values, compares the superposition result with the set multi-parameter coupling threshold, filters the time index segment that exceeds the superposition limit, and generates an over-limit superposition segment set; The abnormal segment statistical extraction submodule marks the time index of the over-limit superposition segment set, counts the occurrence frequency and cumulative concentration of each segment within the continuous monitoring period, integrates and outputs the segments whose occurrence frequency is higher than the set frequency threshold and whose concentration density exceeds the limited interval, establishes a mapping between the corresponding time index and the parameter sequence number, and generates a deformation coupling abnormal segment set.

Images