Thermal fatigue test method and platform

Abstract

The application discloses a thermal fatigue test method and platform, and relates to the field of fatigue test. In the heating stage, the vertex temperature is collected by a temperature measuring instrument, a discrete heating model based on the finite element method and the Fourier heat conduction law is combined to simulate and predict, the power of the high-frequency heating device is updated by rolling optimization and the operation of the high-frequency heating device is controlled; in the cooling stage, the vertex temperature is collected, a discrete cooling model is combined to simulate and predict, the power of the water chiller and the air compressor is updated by rolling optimization and the operation of the water chiller and the air compressor is controlled; in the hot and cold inspection stages, the operation of the ultrasonic device is controlled, the defect features in the ultrasonic signal are extracted based on wavelet decomposition, the detection judgment is performed through the support vector machine, and it is decided whether to terminate the test. The platform simulates the cold and hot change process to realize accurate and rapid temperature control, and improves the reliability of the thermal fatigue test and the defect detection precision.

Description

Technical Field

[0001] This invention relates to the field of fatigue testing, specifically to thermal fatigue testing methods and platforms. Background Technology

[0002] Thermal fatigue testing is a key technology for evaluating the fatigue resistance of materials or components under cyclic temperature changes. It is crucial for ensuring the reliability and durability of high-end equipment in aerospace, energy, and precision machinery industries in complex thermal environments. Accurate thermal fatigue testing provides critical information for material selection and structural design optimization, preventing safety accidents or performance degradation caused by thermal fatigue failure. It plays an irreplaceable role in advanced manufacturing.

[0003] However, traditional thermal fatigue testing methods have technical bottlenecks. The temperature control process lacks accurate simulation of dynamic heat conduction characteristics and cannot optimize heating or cooling power based on real-time temperature data, resulting in a large deviation between the test conditions and actual application scenarios. Summary of the Invention

[0004] This invention addresses the shortcomings of existing technologies by proposing a thermal fatigue testing method and platform to provide rapid temperature control and high-precision defect detection based on accurate temperature simulation of thermal conduction characteristics during thermal fatigue testing.

[0005] The technical solution to achieve the purpose of this invention is as follows:

[0006] The thermal fatigue testing method includes the following steps:

[0007] Get the upper limit temperature With lower limit temperature Heating start-up phase, before initialization The heating power of the first step, in the first step Step-by-step acquisition of thermal temperature vector Combined with the first Step to the first The heating power of the first step is predicted using a discrete heating model to generate the second step. Step-by-step simulated thermal temperature vector The decision on whether to update the number of iterations through rolling optimization. heating power of step And run, until the first The thermal detection phase is initiated step-by-step, in which the discrete heating model is a variational temperature equation derived from a finite element method simulation. Considering the source of action within each unit The stepwise discrete representation of the heat transfer effect generation, the first Step thermal temperature measurement vector include Vertex temperature at each vertex of each cell This represents the total number of steps.

[0008] Ultrasonic testing is performed during the thermal detection phase, and wavelet basis functions are used to analyze the ultrasonic signals. Perform wavelet decomposition and extract and generate defect feature vectors Based on the decision results of the support vector machine, the thermal fatigue test is stopped and the total number of stages is counted or the number of heating stages is updated, and the cooling stage is started.

[0009] Before initialization during the cooling phase The water-cooling power and air-cooling power of the first step, in the first Step-by-step acquisition of cold temperature vector Combined with the first Step to the first The water-cooling power and air-cooling power of the first step are predicted using a discrete cooling model to generate the second step. Step-by-step simulated cold temperature vector The decision on whether to update the number of iterations through rolling optimization. Step water cooling power and air-cooled power And run, until the first The cold detection phase is initiated during the step-by-step process, in which the discrete cooling model is a variable-temperature equation. Considering the source of action within each unit The step-by-step discrete representation of the effects of heat exchange on the transformation process;

[0010] Ultrasonic testing is performed during the cold inspection phase, and wavelet basis functions are used to analyze the ultrasonic signals. Perform wavelet decomposition and extract and generate defect feature vectors Based on the decision results of the support vector machine, the thermal fatigue test is stopped and the total number of stages is counted or the number of cooling stages is updated, and then the heating stage is started.

[0011] Specifically, the variable temperature equation is constructed based on Fourier's law of heat conduction, where temperature... Regarding time Temperature time-varying term With density and specific heat capacity The product of and equals temperature Laplace operator With thermal conductivity The source of the product addition action Among them, temperature Laplace operator equal to temperature The sum of the second-order gradients in the three coordinate axes.

[0012] Furthermore, the preliminary steps for constructing discrete heating and discrete cooling models include:

[0013] Temperature Discretized The element temperature function of the nth element, the th Element temperature function of each element This is a linear interpolation form of the vertex temperature versus volume function for each vertex within the cell. For the first The coordinates of any point within a unit. , The total number of units;

[0014] Temperature is determined using the volume integral method. Replace with the first Element temperature function of each element and the Integrate the volume function of each vertex of each unit, and then rearrange to generate the volume of the first unit. The element matrix equation of the nth element, i.e., the nth element The heat capacity matrix of each unit With partial differential terms Product heating conduction matrix With vertex temperature vector The product equals the load vector With unit action source The accumulation of;

[0015] The first Partial differential terms of each unit Substitute the first Vertex temperature vector of the step And perform time discretization, partial differential terms equal to the Step to the first The change in the vertex temperature vector of a step divided by the duration of a single step. Organize and generate the first Discrete variable temperature equations for each unit.

[0016] Furthermore, based on the finite element method, the sample simulation is divided into... The tetrahedral unit, the first Each unit comprises four vertices. Based on the principle of small-scale approximation, the first... Element temperature function of each element equal to the The linear interpolation result of the vertex temperature and the corresponding volume function of each vertex within each cell, where the... Unit 1 Volume function of vertices For the first Any point within a given unit and the... The tetrahedron formed by the three vertices other than the first vertex and the second vertex The volume ratio of each unit .

[0017] Furthermore, the variable-temperature equation offsets the margin using the volume integral method, thus reducing the first... Unit 1 Volume function of vertices As a weight, it is multiplied by the temperature variation equation and applied to the unit volume. Calculate the volume integral and rearrange it into matrix product form, then generate the first... Unit 1 The equation of the unit vertex matrix of the nth vertex, i.e. the nth vertex The heat capacity matrix of each unit The Middle Line heat capacity vector With partial differential terms Product heating conduction matrix The Middle Linear heat conduction vector With vertex temperature vector The product equals the load vector The Middle Row load elements With unit action source The product of, spliced ​​together The equation of the vertex matrix of the four vertices in the first unit is used to generate the first unit. The element matrix equation of the nth element, where the nth element... Linear heat capacity vector Reflecting the The temperature change of each vertex within a given unit affects the temperature of the first vertex. The heat retention effect at the first vertex, the first Linear heat conduction vector Reflecting the The temperature change of each vertex within a given unit affects the temperature of the first vertex. The effect of vertex temperature change on the first vertex, the second vertex Row load elements Reflecting the Unit action source of each unit For the External influences of each vertex .

[0018] Specifically, the first Element source of action in the discrete variable temperature equation of an element Replace with the first heating power of step With the Heat distribution ratio of each unit The product and the sum of the first Step process noise Organized into the first The unit temperature rise equation for the nth unit, i.e., the nth unit Step-by-step simulation unit thermal temperature vector equal to the The unit matrix of each unit With the Step unit thermal temperature measurement vector The product of the unit decay matrix With the heating power of step The product of the first addition Step process noise , among which, the Heat distribution ratio of each unit equal to the The distance between the center coordinates of each unit and the heating center coordinates is substituted into a pre-fitted Gaussian mixture distribution to generate the unit matrix. Reflecting the Temperature state transition process of each unit, unit decay matrix Reflecting the The unit structure and spatial distance of the first unit cause the... heating power of step Heating attenuation, splicing A discrete heating model is constructed using the unit heating equations for each unit.

[0019] Specifically, the steps for constructing a discrete heating model are basically the same as those for constructing a discrete cooling model, the difference being that the first step... Unit action source of each unit Replace with the first Step water cooling power With the Water cooling distribution ratio of each unit The product plus the first Step air-cooled power With the Air-cooling distribution ratio of each unit The product of, where, the product of, Water cooling distribution ratio of each unit Equal to the total number of units The reciprocal of the air-cooled distribution ratio Then based on the first The distance between the center coordinates of each cell and the center coordinates of the blowing air is substituted into a pre-fitted Gaussian mixture distribution to generate the distribution.

[0020] Furthermore, the decision on whether to perform rolling optimization between the heating and cooling stages includes:

[0021] Based on the initial determination of the first Step to the first The simulation calculation of heating power, water cooling power, or air cooling power in the first step. Step-by-step simulated thermal temperature vector Or simulate cold temperature vector The corresponding subtraction of the upper limit temperature or lower limit temperature And calculate the second norm value as the first Step-by-step simulated thermal temperature difference Or simulate cold temperature difference ;

[0022] If the first Simulated temperature difference of the step If the temperature difference is less than the threshold, based on the initial determination of the first... heating power of step or the Step water cooling power Air-cooled power Run, of which, the first Simulated temperature difference of the step The heating stage or cooling stage respectively represent the first Step-by-step simulated thermal temperature difference Or simulate cold temperature difference ;

[0023] If the first Step-by-step simulated temperature difference If the temperature difference is greater than or equal to the temperature difference threshold, then rolling optimization is performed to minimize the first... Step-by-step simulated thermal temperature difference Or simulate cold temperature difference With the The optimization objective is to determine the power change at each step. A quadratic objective function is established with constraints set, namely the power change at the first step. heating power of step Or water cooling power Air-cooled power All are within the corresponding preset power range;

[0024] Solve the quadratic objective function using a quadratic programming solver and update the result to obtain the... heating power of step Or water cooling power Air-cooled power And run.

[0025] Furthermore, the ultrasound examination during the hot and cold detection phases includes the following steps:

[0026] Scanning samples to acquire ultrasonic signals Wavelet basis functions are used to analyze ultrasonic signals. conduct Layer wavelet decomposition, transforming ultrasonic signals Decomposed into Detail components at different scales of the layer, the detail components reflecting the ultrasound signal High-frequency information;

[0027] Calculate the component energy and component variance of the detail components of each layer and stitch them together to generate a defect feature vector. ;

[0028] Defect feature vector The input is a support vector machine, which is mapped to a high-dimensional space through a kernel function and then subjected to linear binary classification through a hyperplane to determine whether cracks or defects exist.

[0029] If cracks or defects are found, stop the thermal fatigue test and sum the number of heating stages and cooling stages to calculate the total number of stages.

[0030] If there are no cracks or defects and it is the hot inspection stage, then increment the heating stage number by 1 and start the cooling stage;

[0031] If there are no cracks or defects and it is a cold inspection stage, increment the cooling stage number by 1 and start the heating stage.

[0032] A thermal fatigue testing platform for performing the thermal fatigue testing method includes a high-frequency heating device, a water chiller, water-cooled pipes, an air compressor, an ultrasonic device, a temperature measuring instrument, and a central control device.

[0033] A high-frequency heating device heats the bottom of the sample;

[0034] The water chiller delivers cooling water to the water-cooled pipes;

[0035] A water-cooling pipe passes through the sample core;

[0036] An air compressor blows air onto the outer surface of the sample at specific points.

[0037] Ultrasonic devices scan samples to acquire ultrasonic signals ;

[0038] The thermometer measures the vertex temperature of each vertex in the sample;

[0039] The central control equipment executes the thermal fatigue test method to obtain the upper limit temperature. With lower limit temperature During the heating and cooling phases, the vertex temperature of each vertex is collected, and simulation predictions are performed using discrete heating and cooling models respectively. This determines whether to use rolling optimization to update the power of the high-frequency heating device, water chiller, and air compressor at each step and control their operation. During the hot and cold detection phases, the ultrasonic device is controlled to acquire ultrasonic signals. The ultrasonic testing system uses wavelet decomposition and support vector machine to make decisions on whether to terminate the thermal fatigue test.

[0040] Compared with existing technologies, this invention, based on the finite element method and the discrete temperature model system generated by Fourier's heat conduction law, simulates and predicts complex thermal cycles by using discrete heating and cooling models in the heating and cooling stages respectively, combined with real-time collected peak temperature data. It also employs a rolling optimization strategy to dynamically update heating, water cooling, and air cooling power. Simultaneously, in the defect detection stage, it extracts subtle defect features from ultrasonic signals through wavelet decomposition and combines this with support vector machines for intelligent decision-making. This solves the technical problems of large deviations in operating condition simulation and insufficient defect feature recognition capabilities caused by coarse temperature control in traditional testing, significantly improving the reliability and defect detection accuracy of thermal fatigue testing. Attached Figure Description

[0041] Figure 1 This is a flowchart of the heating and thermal testing stages in the thermal fatigue testing method.

[0042] Figure 2 This is a flowchart of the cooling and cold inspection stages in the thermal fatigue testing method.

[0043] Figure 3 A flowchart for deciding whether to perform rolling optimization between the heating and cooling stages;

[0044] Figure 4 The flowchart shows the ultrasound testing process during the hot and cold testing phases.

[0045] Figure 5 This is a schematic diagram of a thermal fatigue testing platform.

[0046] Reference numerals in the attached diagram: 1. High-frequency heating device; 2. Water chiller; 3. Water-cooled pipe; 4. Air compressor; 5. Ultrasonic device; 6. Thermometer; 7. Central control equipment. Detailed Implementation

[0047] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0048] Example 1:

[0049] like Figure 1 and Figure 2 As shown, this invention discloses a thermal fatigue testing method, comprising the following steps:

[0050] Get the upper limit temperature With lower limit temperature Initiate the heating phase, before initialization during the heating phase. The heating power of the first step, in the first step The thermal vector is acquired by a thermometer. Combined with the first Step to the first The heating power of the first step is predicted step by step using a discrete heating model to generate the heating power of the second step. Step-by-step simulated thermal temperature vector and the upper limit temperature Comparison calculation Step-by-step simulated thermal temperature difference The decision on whether to update the number of iterations through rolling optimization. heating power of step And adjust the operation of the high-frequency heating device until the first... The thermal detection phase is initiated step-by-step, in which the discrete heating model is a variable-temperature equation in... Considering external sources within each unit The stepwise discrete representation of the transformation of heat transfer effects The sample generation is based on the finite element method simulation. Step thermal temperature measurement vector include Vertex temperature at each vertex of each cell This represents the total number of steps.

[0051] During the thermal inspection phase, based on the flaw detection frequency and flaw detection power Adjusting and operating the ultrasonic device for ultrasonic testing, using wavelet basis functions to analyze the ultrasonic signal. conduct Layer wavelet decomposition and extraction to generate defect feature vectors Based on the decision results of the support vector machine, the thermal fatigue test is stopped and the total number of stages is counted or the number of heating stages is updated and the cooling stage is started.

[0052] Before initialization during the cooling phase The water-cooling power and air-cooling power of the first step, in the first The cold temperature vector is acquired by a temperature measuring instrument. Combined with the first Step to the first The water-cooling power and air-cooling power of the first step are generated using a discrete cooling model. Step-by-step simulated cold temperature vector and the lower limit temperature Comparison calculation Step-by-step simulated cold temperature difference The decision on whether to update the number of iterations through rolling optimization. Step water cooling power and air-cooled power And adjust the operation of the water chiller and air compressor until the first The cold detection phase is initiated during the step-by-step process, in which the discrete cooling model is a variable-temperature equation. Considering external sources within each unit The step-by-step discrete representation of the effects of heat exchange on the transformation process;

[0053] During the cold inspection phase, based on the flaw detection frequency and flaw detection power Adjusting and operating the ultrasonic device for ultrasonic testing, using wavelet basis functions to analyze the ultrasonic signal. conduct Layer wavelet decomposition and extraction to generate defect feature vectors Based on the decision results of the support vector machine, the thermal fatigue test is stopped, the total number of stages is counted or the number of cooling stages is updated, and the heating stage is started.

[0054] Specifically, the temperature variation equation is constructed in partial differential form based on Fourier's law of heat conduction, and the temperature of the sample... Regarding time The rate of change depends on internal heat conduction and external heat sources. The synergistic effects are as follows:

[0055] ,

[0056] in, For temperature time-varying terms, This is the thermal conductivity term, reflecting the spatial transfer of heat within the sample. , and These are thermal conductivity, density, and specific heat capacity, which are directly related to the sample material. For temperature The Laplace operator is equal to temperature. The sum of the second-order gradients in the three coordinate axes of a Cartesian coordinate system. External action terms reflect the source of action. For the temperature of the sample The direct impact.

[0057] Furthermore, the construction of discrete heating models and discrete cooling models share the same preliminary steps, including:

[0058] Temperature of the sample Discretized The element temperature function of the nth element, the th Unit temperature of each unit Based on the The volume function at each vertex within a cell is transformed into a linear interpolated form of the cell temperature function. ,in, For the first The coordinates of any point within a unit. , The total number of units;

[0059] The temperature in the variable temperature equation is obtained by using the volume integral method. Replace with the first Element temperature function of each element and the Integrate the volume function of each vertex of each unit, and then rearrange to generate the volume of the first unit. The element matrix equations for each element are as follows:

[0060] ,

[0061] in, , , and The first The heat capacity matrix, vertex temperature vector, heat conduction matrix, and load vector of each unit. For the first The unit's source of action;

[0062] The first Partial differential terms in the unit matrix equation of each unit In the Step-by-step time discretization, vertex temperature vector In essence, it is the first Vertex temperature vector of the step Partial differential terms equal to the Vertex temperature vector of the step With the Vertex temperature vector of the step Change divided by single step duration Organize and generate the first The discrete temperature equations for each unit are as follows:

[0063] ,

[0064] in, For the first The unit matrix of the first unit comprehensively reflects the first unit. Temperature state transition process of each unit For the first The unit in the first The unit action source of the step, the first The discrete variable-temperature equations for each unit are presented in a step-by-step discrete form. The process of the vertex temperature of each unit changing with the number of steps is used to progressively predict the temperature of the first unit during the heating and cooling phases. Step-by-step simulated thermal temperature vector and the Step-by-step simulated cold temperature vector Provide a pre-built model.

[0065] Furthermore, based on the finite element method, the sample simulation is divided into... Tetrahedral units of the same specifications, the first Each unit comprises four vertices. Based on the principle of small-scale approximation, when the unit's geometric size is small, the temperature at any point within the unit follows a linear distribution in space, i.e., the... Element temperature function of each element equal to the The linear interpolation result of the vertex temperature and the corresponding volume function of each vertex within each cell, the i-th Element temperature function of each element Specifically as follows:

[0066] ,

[0067] in, and The first Unit 1 The volume function and vertex temperature of the nth vertex, Volume function of vertices Defined as the first Any point within a given unit and the... The tetrahedron formed by the three vertices other than the first vertex and the second vertex The volume ratio of each unit .

[0068] Furthermore, since the variable-temperature equation is in partial differential form, direct solution will produce a margin. The volume integration method uses the volume function of each vertex in the element as a weight and performs volume integration on the variable-temperature equation substituted with the element temperature function within the element, thus offsetting the margin through a weighted sum. Unit 1 Volume function of vertices The volume integral used as a weight is as follows:

[0069] ,

[0070] in, The volume of each element is the same because the elements divided by the finite element method have the same dimensions.

[0071] Calculate the time-varying term of temperature and the first Unit 1 Volume function of vertices The integral, rearranged into a matrix product form, is as follows:

[0072] ,

[0073] in, For the first Unit 1 Volume function of vertices, and The first The heat capacity matrix of each unit The Middle Line number The heat capacity element of the column and the first heat capacity vector, heat capacity element Reflecting the Within the unit, the first The temperature change of the vertex at the 1st vertex affects the 2nd vertex. The impact of heat retention at each vertex Represents the transpose of a vector or matrix. ;

[0074] Calculate the heat conduction term and the first Unit 1 Volume function of vertices The integral, rearranged into a matrix product form, is as follows:

[0075] ,

[0076] in, For the first Unit 1 The Hamiltonian operator for vertices is equal to the volume function. The sum of the first-order gradients in the directions of the three coordinate axes in a Cartesian coordinate system. and The first Heat conduction matrix of each unit The Middle Line number The heat conduction elements of the column and the first Linear heat conduction vector, heat conduction element Reflecting the Within the unit, the first The temperature change of the vertex at the 1st vertex affects the 2nd vertex. The effect of vertex temperature changes at each vertex;

[0077] Calculate the external action term and the first Unit 1 Volume function of vertices The integral, rearranged into a matrix product form, is as follows:

[0078] ,

[0079] in, For the first Load vector of each unit The Middle The load elements of the row reflect the first Unit action source of each unit For the External influences of each vertex;

[0080] The first Unit 1 Volume function of vertices The volume integral, used as a weight, is rearranged based on the matrix product of the integrals of the time-varying temperature term, the heat conduction term, and the external action term to generate the first... Unit 1 The equation for the unit vertex matrix of vertices is as follows:

[0081] ,

[0082] The first The equations of the vertex matrix of each unit with 4 vertices are concatenated row by row to generate the first unit. The unit matrix equation of each unit .

[0083] Furthermore, constructing a discrete heating model also includes the following steps:

[0084] The first Unit action source of each unit Replace with the first heating power of step With the Heat distribution ratio of each unit The product of, where, the product of, Heat distribution ratio of each unit equal to the The distance between the unit center coordinates and the heating center coordinates of the high-frequency heating device is substituted into the Gaussian mixture distribution to generate the distribution, which is established by pre-fitting through experiments.

[0085] During the heating phase, in the first Based on the discrete temperature variation equations of the first unit, considering the noise interference introduced by the temperature measuring instrument, it is transformed into the first... The unit temperature rise equation for the nth unit, in the nth unit Step unit thermal temperature measurement vector Given the known information, predict the first... Step-by-step simulation unit thermal temperature vector The details are as follows:

[0086] ,

[0087] in, and Each includes the first The four vertices of the unit are in the... The simulated peak temperature of the first step and the first step The peak temperature of the first step, Step unit thermal temperature measurement vector Upon reaching the The temperature was collected during the step, and before reaching the first step... Take the first step directly. Step-by-step simulation unit thermal temperature vector , For the first The cell attenuation matrix of the nth cell reflects the cell attenuation matrix of the nth cell. The unit structure and spatial distance of the first unit cause the... heating power of step Heating attenuation, For the first The superimposed process noise reflects the prediction error of the equation and the acquisition error caused by environmental noise interference;

[0088] Will The unit heating equations for each element are concatenated row by row to construct a discrete heating model, as follows:

[0089] ,

[0090] in, , and The first The simulated thermal temperature vector of the first step, the first The thermal measurement vector and attenuation matrix of the step.

[0091] Furthermore, constructing a discrete cooling model also includes the following steps:

[0092] The first Unit action source of each unit Replace with the first Step water cooling power With the Water cooling distribution ratio of each unit The product plus the first Step air-cooled power With the Air-cooling distribution ratio of each unit The product of, where, the product of, Water cooling distribution ratio of each unit Equal to the total number of units The reciprocal of the value is because the coolant from the water chiller passes evenly through the sample core, thus affecting the sample. Each unit has the same function, but the air-cooling distribution ratio is different. Then based on the first The distance between the center coordinates of each unit and the air compressor's blowing center is substituted into a Gaussian mixture distribution to generate the distribution, which is established through pre-fitting experiments.

[0093] During the cooling phase, in the first Based on the discrete temperature variation equations of the first unit, considering the noise interference introduced by the temperature measuring instrument, it is transformed into the first... The unit cooling equation for the nth unit, in the nth unit Step unit cold temperature vector Given the known information, predict the first... Step-by-step simulation unit cold temperature vector The details are as follows:

[0094] ,

[0095] in, and Each includes the first The four vertices of the unit are in the... The simulated peak temperature of the first step and the first step The peak temperature of the first step, Step unit cold temperature vector Upon reaching the The temperature was collected during the step, and before reaching the first step... Take the first step directly. Step-by-step simulation unit cold temperature vector , and The first The unit water-cooled attenuation matrix and the unit air-cooled attenuation matrix of the first unit reflect the first unit's... The unit structure and spatial distance of the first unit cause the... Step water cooling power With air-cooled power Cooling attenuation, For the first The unit cooling attenuation matrix of each unit. For the first The cooling power vector of the step;

[0096] Will The unit cooling equations of each element are concatenated row by row to construct a discrete cooling model, as follows:

[0097] ,

[0098] in, , and The first The simulated cold temperature vector of the first step, the first The cold temperature measurement vector and attenuation matrix of the step.

[0099] like Figure 3 As shown, furthermore, the decision on whether to perform rolling optimization in the heating and cooling stages has the same processing steps, including:

[0100] Based on the initial determination of the first Step to the first The power simulation calculation of the first step yields the... Step-by-step simulated temperature measurement vector The power during the heating or cooling stage is either the heating power or the water cooling power and the air cooling power, respectively. Step-by-step simulated temperature measurement vector The heating stage or cooling stage respectively represent the first Step-by-step simulated thermal temperature vector Or simulate cold temperature vector ;

[0101] The first Step-by-step simulated temperature measurement vector Subtract the limit temperature And calculate the second norm value as the first Step-by-step simulated temperature difference Limit temperature The heating stage and cooling stage represent the upper limit temperature, respectively. or lower limit temperature , No. Step-by-step simulated temperature difference The heating stage or cooling stage respectively represent the first Step-by-step simulated thermal temperature difference Or simulate cold temperature difference ;

[0102] If the first Step-by-step simulated temperature difference If the temperature difference is less than the threshold, then the initial determination of the first... Step power The temperature control device is operated, and the temperature control device is a high-frequency heating device or a water chiller and an air compressor, respectively, during the heating or cooling stage;

[0103] If the first Step-by-step simulated temperature difference If the temperature difference is greater than or equal to the temperature difference threshold, then rolling optimization is performed. The optimization objective of rolling optimization is to minimize the first... Step-by-step simulated temperature difference and the The rate of change of power in step 1, the first step The rate of change of power in step 1 is equal to that in step 2. Step power Subtract the first The final step determined in the process Step power ;

[0104] The first Step power As an optimization variable, the first The rate of change of power at step 1 is expressed in the form of a regularization term with the first step 2. Step-by-step simulated temperature difference Summing is performed to construct a quadratic objective function, and a power constraint is set for the quadratic objective function, namely the power of the first phase during the heating stage. heating power of step Always within the preset heating power range, during the cooling phase Step water cooling power With air-cooled power They are respectively located within the preset water-cooling power range and the preset air-cooling power range;

[0105] Solve the quadratic objective function using a quadratic programming solver and update the result to obtain the... Step power It also operates a temperature control device, and the quadratic programming solvers include CPLEX and GUROBI.

[0106] like Figure 4 As shown, the steps for ultrasound testing in the hot and cold testing phases are exactly the same, the only difference being the specific operations taken based on the judgment result. Ultrasound testing includes the following steps:

[0107] Based on flaw detection frequency and flaw detection power Adjust the operation of the ultrasonic device to scan the sample and acquire the ultrasonic signal of the sample. ;

[0108] Wavelet basis functions are used to analyze ultrasonic signals conduct Layer wavelet decomposition, transforming ultrasonic signals Decomposed to Generate frequency subbands at different levels Detail components at different scales of the layer and the first Approximate components of the layer , Layer wavelet decomposition is as follows:

[0109] ,

[0110] in, For the first The detailed components of the layer reflect the ultrasonic signal. The high-frequency information, the reflected signal generated by the damage, passes through the detail components located at high frequencies, the first Approximate components of the layer This reflects the ultrasound signal. The low-frequency trend;

[0111] Calculate the first Layer detail components Component energy and component variance The details are as follows:

[0112] ,

[0113] ,

[0114] in, and These refer to the detection time range and detection duration, respectively. For the first Layer detail components Within the detection time range Average detail component within, ;

[0115] Will The component energy and component variance of the detailed components of the layer are concatenated column by column to generate the defect feature vector. ;

[0116] Defect feature vector The input is a support vector machine, which is mapped to a high-dimensional space through a kernel function and then subjected to linear binary classification through a hyperplane to determine whether cracks or defects exist.

[0117] If cracks or defects are found, immediately stop the thermal fatigue test and sum the number of heating stages and cooling stages to calculate the total number of stages.

[0118] If there are no cracks or defects and it is the hot inspection stage, then increment the heating stage number by 1 and start the cooling stage;

[0119] If there are no cracks or defects and it is a cold inspection stage, increment the cooling stage number by 1 and start the heating stage.

[0120] Example 2:

[0121] like Figure 5 As shown, the present invention also discloses a thermal fatigue testing platform for performing the thermal fatigue testing method, including a high-frequency heating device 1, a water chiller 2, a water-cooled pipe 3, an air compressor 4, an ultrasonic device 5, a thermometer 6, and a central control device 7.

[0122] The high-frequency heating device 1 is controlled by the central control equipment 7 and heats the bottom of the sample through a high-frequency induction coil;

[0123] The water chiller 2 is controlled by the central control equipment 7 and delivers cooling water with a pH value of 6 to 8, a hardness of less than or equal to 10 degrees and a resistivity of greater than 2500Ω / cm into the water cooling pipe 3. The temperature of the cooling water is fixed.

[0124] Water-cooled pipe 3 passes through the sample core to achieve uniform cooling of the sample core by cooling water;

[0125] The air compressor 4 is controlled by the central control equipment 7, which compresses the air and blows it onto the outer surface of the sample through a high-pressure air pipe.

[0126] The ultrasonic device 5 is controlled by the central control equipment 7, using a flaw detection frequency. and flaw detection power The sample was ultrasonically scanned to acquire ultrasonic signals. ;

[0127] The thermometer 6 is controlled by the central control device 7 and measures the temperature of the sample based on the principle of infrared radiation. Vertex temperature of each vertex in each unit;

[0128] The central control device 7 executes the thermal fatigue test method to obtain the upper limit temperature. With lower limit temperature The system cycles through heating, hot detection, cooling, and cold detection phases. During the heating and cooling phases, the temperature at each peak is collected by a thermometer 6, and simulations are performed using discrete heating and cooling models. The system autonomously decides whether to use rolling optimization to update the power of the high-frequency heating device 1, water chiller 2, and air compressor 4 at each step and controls their operation accordingly. During the hot and cold detection phases, the ultrasonic device 5 is controlled to obtain ultrasonic signals. The ultrasonic testing system uses wavelet decomposition and support vector machine to make decisions on whether to stop the thermal fatigue test.

[0129] This invention discloses a thermal fatigue testing method and platform. During the heating phase, the peak temperature is collected using a thermometer, and simulation prediction is performed using a discrete heating model constructed based on the finite element method and Fourier's law of heat conduction. The power of the high-frequency heating device is continuously optimized and its operation controlled. During the cooling phase, the peak temperature is collected, and simulation prediction is performed using a discrete cooling model. The power of the water chiller and air compressor is continuously optimized and their operation controlled. During the hot and cold inspection phases, the ultrasonic device is controlled, and defect features in the ultrasonic signal are extracted based on wavelet decomposition. A support vector machine is used for detection and decision-making, autonomously deciding whether to terminate the test. This platform simulates the hot and cold transformation process to achieve precise and rapid temperature control, improving the reliability of thermal fatigue testing and the accuracy of defect detection.

[0130] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for testing thermal fatigue, characterized in that, Includes the following steps: Heating start-up phase, before initialization The heating power of the first step, in the first step The first step involves acquiring thermal temperature vectors and using a discrete heating model to predict and generate the second step. The simulated thermal temperature vector of the first step determines whether to update the second step through rolling optimization. The heating power is adjusted and the circuit is run until the next step. The thermal detection phase is initiated during the step. This represents the total number of steps. During the thermal inspection phase, ultrasonic signals are acquired and defect feature vectors are generated by wavelet decomposition. Based on the decision results of the support vector machine, the test is stopped and the total number of phases is counted or the number of heating phases is updated, and the cooling phase is started. Before initialization The water-cooling power and air-cooling power of the first step, in the first The first step involves acquiring cold temperature vectors and using a discrete cooling model to predict and generate the second... The simulated cold temperature vector of the first step determines whether to update the second step through rolling optimization. Step water cooling power and air-cooled power And run, until the... The cold detection phase is initiated during the step; During the cold inspection phase, ultrasonic signals are acquired and defect feature vectors are generated by wavelet decomposition. Based on the decision results of the support vector machine, the test is stopped and the total number of phases is counted or the number of cooling phases is updated, and then the heating phase is started. The discrete heating model and discrete cooling model are generated at different stages based on the elements divided by the finite element method and the variable temperature equation constructed based on Fourier's law of heat conduction. The variable temperature equation is constructed based on Fourier's law of heat conduction. The product of the temperature time-varying term with respect to time and the density and specific heat capacity is equal to the product of the Laplace operator of temperature and the thermal conductivity plus the source. The Laplace operator of temperature is equal to the sum of the second-order gradients of temperature in the three coordinate axis directions. The preliminary steps for constructing discrete heating and discrete cooling models include: discretizing the temperature into... The element temperature function for each element is a linear interpolation of the vertex temperature and volume function at each vertex within the element. The total number of elements is given. The temperature is replaced by the element temperature function of each element using the volume integration method, and then integrated with the volume function of each vertex within the element. This process generates the element matrix equation for each element, where the product of the heat capacity matrix and partial differential terms, and the product of the heating conduction matrix and the vertex temperature vector, equals the product of the load vector and the element's source of action. Substituting the partial differential terms of each element into the... The vertex temperature vector of the step is discretized over time, and the partial differential term is equal to the first step. Step to the first The change in the vertex temperature vector of each step is divided by the single-step duration, and the discrete variable temperature equations for each unit are generated. Constructing a discrete heating model also includes the following steps: replacing the element source in the discrete temperature change equation of each element with the element source of the first element. The product of the heating power of the first step and the heat distribution ratio of each unit, and then superimposed on the first step. The process noise of the first step is summarized into the unit temperature rise equation for each unit, i.e., the first step... The simulated element thermal vector of the step is equal to the element matrix of each element and the first element. The product of the unit thermal temperature measurement vectors of the first step plus the unit attenuation matrix and the first step The product of the heating power of the first step plus the second step The process noise of the first step is generated by substituting the heat distribution ratio of each unit into the distance between the unit center coordinates and the heating center coordinates of each unit into a pre-fitted Gaussian mixture distribution. The unit matrix reflects the temperature state transition process of each unit, and the unit attenuation matrix reflects the unit structure and the noise caused by spatial distance. The heating power attenuation of the step, splicing Construct a discrete heating model using the unit heating equations of each unit; The difference between constructing a discrete cooling model and constructing a discrete heating model lies in replacing the element source of each element with the element source of the first element. The product of the water-cooling power of the first step and the water-cooling allocation ratio of each unit plus the second step The product of the air-cooling power of the step and the air-cooling allocation ratio of each unit, where the water-cooling allocation ratio of each unit is equal to the total number of units. The reciprocal of the given value, and the air-cooling distribution ratio is generated by substituting the distance between the unit center coordinates and the air blowing center coordinates of each unit into a pre-fitted Gaussian mixture distribution.

2. The thermal fatigue testing method as described in claim 1, characterized in that, Based on the finite element method, the sample simulation is divided into... Each tetrahedral element has four vertices. Based on the principle of small-scale approximation, the element temperature function of each element is equal to the linear interpolation result of the vertex temperature of each vertex in the element and the corresponding volume function. The volume function of a single vertex in the element is the ratio of the volume of the tetrahedron formed by any point in the element and the three vertices other than the single vertex to the volume of the element.

3. The thermal fatigue testing method as described in claim 1, characterized in that, The volume integral method is used to offset the residual of the variable temperature equation, and to integrate the volume integral within each unit. The volume function of each vertex is used as a weight to multiply with the temperature variation equation, and the volume integral is calculated within the unit volume. This is then rearranged into a matrix product form to generate the volume function of each unit. The equation of the unit vertex matrix for the nth vertex, that is, the equation of the heat capacity matrix of each unit for the nth vertex. The product of the heat capacity vector and the partial differential term in the heating conduction matrix is ​​the first... The product of the heat conduction vector and the vertex temperature vector in the row is equal to the product of the load vector in the first row. Load elements and element action sources in the row The product of these components is used to concatenate the unit vertex matrix equations of the four vertices of each unit, generating the unit matrix equation for each unit. The product of these components and the concatenation of the product of these components are then used to generate the unit matrix equation for each unit. The heat capacity vector of a row reflects the effect of the vertex temperature change at each vertex within each cell on the first vertex. The heat retention effect at the first vertex, the first The heat conduction vector of a row reflects the effect of the vertex temperature change at each vertex within each cell on the first vertex. The effect of vertex temperature change on the first vertex, the second vertex The load elements of the row reflect the element action source of each element on the first element. External influences of each vertex.

4. The thermal fatigue testing method as described in claim 1, characterized in that, Whether to perform rolling optimization for the heating and cooling phases includes: Based on the initial determination of the first Step to the first The simulation calculation of heating power, water cooling power, or air cooling power in the first step. The simulated thermal or cold temperature vector of the step is subtracted from the upper or lower limit temperature, and the second norm value is calculated as the first value. The simulated thermal temperature difference or simulated cold temperature difference; If the first If the simulated temperature difference in step one is less than the temperature difference threshold, then based on the initial determination of the first step... heating power of step or the first The water-cooled power and air-cooled power are operated in stages, where the simulated temperature difference in the heating stage or cooling stage represents the simulated hot temperature difference or the simulated cold temperature difference, respectively. If the first If the simulated temperature difference in the first step is greater than or equal to the temperature difference threshold, then rolling optimization is performed to minimize the first step. The simulated thermal temperature difference or simulated cold temperature difference in the first step and the first The optimization objective is to determine the power change at each step. A quadratic objective function is established with constraints set, namely the power change at the first step. The heating power, water cooling power, or air cooling power of each step are all within the corresponding preset power range; Solve the quadratic objective function using a quadratic programming solver and update the result to obtain the... The heating power, water cooling power, or air cooling power of the step are adjusted and the system is in operation.

5. The thermal fatigue testing method as described in claim 1, characterized in that, Ultrasonic testing in both the hot and cold testing phases includes the following steps: The ultrasound signal is decomposed into wavelet basis functions. Layer detail components; Calculate the component energy and component variance of the detail components of each layer and splice them to generate a defect feature vector; The defect feature vector is input into the support vector machine, and linear binary classification is performed by mapping through the kernel function and using the hyperplane to determine whether a crack defect exists. If cracks or defects are found, stop the thermal fatigue test and sum the number of heating stages and cooling stages to calculate the total number of stages. If there are no cracks or defects and it is the hot inspection stage, then increment the heating stage number by 1 and start the cooling stage; If there are no cracks or defects and it is a cold inspection stage, increment the cooling stage number by 1 and start the heating stage.

6. A thermal fatigue testing platform, characterized in that, The system includes a central control device that executes the thermal fatigue testing method according to any one of claims 1-5, obtains the upper limit temperature and the lower limit temperature, collects the peak temperature of each peak during the heating and cooling stages, performs simulation predictions by combining discrete heating models and discrete cooling models respectively, decides whether to obtain the power of the high-frequency heating device, water chiller and air compressor at each step through rolling optimization updates and controls their operation, controls the operation of the ultrasonic device to obtain ultrasonic signals during the hot detection and cold detection stages, performs ultrasonic detection judgment based on wavelet decomposition and support vector machine and autonomously decides whether to stop the thermal fatigue test.

Images