Method for evaluating strain aging degree of welded steel structure

Abstract

The present application relates to a kind of detection methods for evaluating the strain aging degree of welded steel structure, the present application carries out tensile test to the welded joint test piece of different pre-strain, different aging treatment, and utilizes AE detection technology and IRT detection technology to carry out real-time monitoring, obtains AE time domain signal and temperature time history, obtains AE local energy ratio, AE damage parameter, temperature rise rate by post-processing and determines it as input layer, and strain aging damage index is output layer, to establish neural network model.For the sample to be detected, AE detection and IRT detection are carried out on the tensile process, and the characteristic parameters processed after the characteristic parameters are input into the neural network model established, and the strain aging degree can be evaluated.The method of the present application is not limited by the type of welding defects, sampling position, sampling quantity and the like, and the two detection techniques are mutually compensated in mechanical energy dissipation, improving the accuracy of strain aging degree grade evaluation.

Description

Testing methods for assessing the degree of strain aging in welded steel structures Technical Field

[0001] This invention relates to a safety evaluation technology for welded steel structures, specifically a testing method for assessing the degree of strain aging in welded steel structures. Background Technology

[0002] The production of pipeline steel involves numerous complex processes. Among these, processes such as spiral forming, welding, and corrosion protection can alter the material's properties to varying degrees. During spiral forming and internal / external corrosion protection (100–200℃), mechanical energy is converted and dissipated. The consumed mechanical energy is converted into heat and strain energy, which is dissipated as elastic waves via acoustic emission. This pre-deformed welded pipeline steel undergoes varying degrees of strain aging, resulting in reduced toughness and plasticity, shortened service life, and potentially jeopardizing energy security and public safety. Therefore, real-time monitoring of the strain aging degree is necessary for safety assessment of welded steel structures.

[0003] Researchers in various fields of materials and safety testing engineering have conducted extensive studies on the strain aging of steel. However, the research methods are relatively concentrated, analyzing the mechanical response through tensile and impact tests, and assessing the degree of strain aging of the material using parameters such as yield strength, elongation at break, and impact energy. For example, the invention patent application No. 201510926271.4, "A Method for Evaluating the Degree of Strain Aging of Pipeline Steel Pipes," proposes a method for evaluating the degree of strain aging of pipeline steel pipes. This method calculates the product of the decrease in impact energy and the increase in yield strength as the strain aging coefficient δ of the steel to evaluate the degree of strain aging after pipeline steel pipe construction. The invention patent application No. 201611101028.X, "A Non-destructive Testing and Evaluation Method for Strain Aging Embrittlement of 20G Steel," proposes a non-destructive testing and evaluation method for strain aging embrittlement of 20G steel. This method detects the strain using a magnetization characteristic measurement device, and then judges the state and nitrogen content of the tested equipment to assess whether strain aging embrittlement has occurred and the degree of strain aging embrittlement. The invention patent with patent application number 202211318019.1, entitled "A Non-destructive Testing Method and System for High-strength Steel Considering the Influence of Strain Aging," proposes a relatively novel method. This method acquires AE signals through a lead-breaking experiment and constructs an AE characteristic parameter library to facilitate strain aging assessment of other damaged steel structures. These three inventions only target the assessment of strain aging degree of unwelded specimens. The assessment effectiveness of the first two inventions is affected by factors such as sampling location and quantity, and the assessment process does not meet requirements for economy and repeatability. The AE signal from the lead-breaking experiment in the third invention is significantly affected by the length, angle, and location of the broken lead, requiring strict operational procedures; even minor oversights can lead to significant errors in the assessment results. Summary of the Invention

[0004] The purpose of this invention is to provide a testing method for evaluating the degree of strain aging of welded steel structures, so as to solve the problem that existing steel structure strain aging testing methods are limited by factors such as sampling location and sampling quantity, and do not meet the requirements of economy and repeatability.

[0005] The present invention is implemented as follows: a testing method for evaluating the strain aging degree of welded steel structures, comprising the following steps.

[0006] a. First, two steel plates from the same batch are butted together by hand and then cut into several specimens of the same shape, with the weld located in the middle of the specimen.

[0007] b. The prepared specimens are pre-stretched to achieve different levels of pre-strain, and the pre-stretched specimens are subjected to artificial aging treatment at different temperatures.

[0008] c. The artificially aged specimens were subjected to uniaxial tensile tests, and real-time monitoring was performed using AE detection technology and IRT detection technology to obtain the time-domain amplitude diagram of the AE signal and the temperature-time history diagram.

[0009] d. Perform a Fast Fourier Transform (FFT) on the AE time-domain signal acquired during the elastoplastic stage, and after post-processing, obtain the local energy ratio (PP) of a specific frequency band in the AE spectrum, and define the AE damage parameter DP.

[0010]

[0011] Wherein, Peak frequency is the peak frequency of the AE signal.

[0012] e. Calculate the temperature change rate ΔW during the strain hardening stage of the tensile process based on the temperature-time history diagram.

[0013] f. Define the strain-aging damage index DI.

[0014] DI = Pre strain Aging temperature

[0015] Among them, Pre strain The term "pre-strain" refers to the pre-strain condition, and "aging temperature" refers to the aging temperature.

[0016] g. Establish a neural network model, train the feedforward network using the backpropagation algorithm, and use the local energy ratio PP of a specific frequency band of the post-processed AE spectrum, the AE damage parameter DP, and the temperature change rate ΔW during the strain hardening stage as the input layer of the network, and the strain aging damage index DI as the output layer.

[0017] h. Obtain the test specimen and monitor the tensile testing process of the test specimen in real time using AE detection technology and IRT detection technology. Substitute the local energy ratio PP, AE damage parameter DP, and temperature rise rate ΔW of the strain hardening stage in a specific frequency band of the AE spectrum into the established neural network model to obtain the predicted strain aging damage index DI.

[0018] In step d, typical AE time-domain signals are selected from the AE detection system, and the time-domain waveform signals are converted into single-column amplitude data. For each signal, a fast Fourier transform is performed to obtain a spectrum diagram. Six frequency band intervals are defined, and the local energy ratio PP of a specific frequency band interval is calculated.

[0019] Find the local energy percentage PP of the AE signal in the 250kHz to 500kHz range.

[0020]

[0021] in, It is a function obtained by performing a Fast Fourier Transform on the AE signal.

[0022] During the training of the feedforward network, the variables are normalized so that all variables are in the range of 0 to 1.

[0023] In step b, the specimen is uniaxially stretched at room temperature to exceed the material yield point to achieve different levels of pre-strain, namely 6%, 10%, and 14%.

[0024] In step b, the pre-stretched specimen is placed in a high-temperature heating furnace and heated to 100°C, 150°C and 200°C respectively. After holding at these temperatures for 120 minutes, the specimen is cooled to room temperature in the furnace and left to stand for 48 hours.

[0025] This invention provides a testing method for assessing the degree of strain aging in welded steel structures. It employs both AE (Adaptive Aesthetic) and IRT (Integrated Thermal Reduction) testing technologies. IRT technology directly reflects the thermodynamics of metal plastic flow and strain localization processes, while AE technology offers ultra-high temporal resolution in characterizing the rapid release of elastic waves. The two technologies can compensate for each other when reflecting mechanical energy dissipation.

[0026] This invention involves tensile testing of welded joint specimens with different pre-strain and aging treatments, and real-time monitoring using AE (Advanced Electrode Spectrometry) and IRT (Intensity Reduction Test) technologies. This yields time-domain amplitude maps of the AE signals and temperature-time history maps, allowing for the determination of the input and output layers and the establishment of a neural network model. For each specimen under test, AE and IRT tests are performed, and the corresponding parameters are input into the established neural network model to evaluate the degree of strain aging. This method is not limited by the type of welding defect, sampling location, or number of samples. Furthermore, the two testing technologies compensate for each other in terms of mechanical energy dissipation, improving the accuracy of strain aging degree evaluation. Attached Figure Description

[0027] Figure 1 is a schematic diagram of the specimen of the present invention.

[0028] Figure 2 shows the stress-strain curves of the welded joint specimens that have undergone strain aging. Figure 2(a) shows the stress-strain curve of the specimen aged at 100℃, Figure 2(b) shows the stress-strain curve of the specimen aged at 150℃, and Figure 2(c) shows the stress-strain curve of the specimen aged at 200℃.

[0029] Figure 3 is a flowchart of the calculation of PP and DP values ​​of the present invention. Figure 3(a) is a time-domain amplitude diagram of the AE signal, Figure 3(b) is the time-domain amplitude data, Figure 3(c) is a frequency-domain amplitude diagram of the AE signal, and Figure 3(d) is the DP value of this AE signal.

[0030] Figure 4 shows the temperature change curves during the tensile process. Figure 4(a) shows the temperature-time curve of the S10-100 specimen, Figure 4(b) shows the temperature-time curve of the S10-150 specimen, and Figure 4(c) shows the temperature-time curve of the S10-200 specimen.

[0031] Figure 5 is a schematic diagram of the fully connected neural network of the present invention. Detailed Implementation

[0032] This invention is a detection method for assessing the degree of strain aging in welded steel structures. It employs acoustic emission (AE) and near-surface infrared thermal imaging (IRT) detection technologies. To provide a more accurate safety assessment of the structure, this invention establishes a fully connected neural network. The neural network model is trained using the local energy value PP of a specific frequency band of the AE spectrum during the elastoplastic stage, the AE damage parameter DP, the temperature change rate ΔW during the strain hardening stage, and the strain aging damage index DI. The results can be obtained by coupling the parameters of the two detection technologies to obtain the strain aging damage index.

[0033] Specifically, it includes the following steps.

[0034] a. First, two steel plates from the same batch are butted together by hand and then cut into several specimens of the same shape, with the weld located in the middle of the specimen.

[0035] After butt welding the steel plates, several specimens as shown in Figure 1 are fabricated using laser cutting. The specimens are "dog-bone" shaped, wider at both ends and narrower in the middle, with the weld located in the center. A sufficient number of specimens should be prepared to accommodate different combinations of pre-strain during pre-tensioning and different aging temperatures during artificial aging.

[0036] Specifically, the specimen dimensions are 222mm in length and 39mm in width. The welding process requires strict control of factors such as welding speed, welding current, and welding temperature. The fabricated welded joint specimens are inspected using radiographic testing to detect various defects, including slag inclusions, porosity, and incomplete penetration.

[0037] b. The prepared specimens are pre-stretched to achieve different levels of pre-strain, and the pre-stretched specimens are subjected to artificial aging treatment at different temperatures.

[0038] The specimens were uniaxially stretched to exceed the material yield point at room temperature to achieve different levels of pre-strain, namely 6%, 10%, and 14%.

[0039] The pre-stretched specimens were then placed in a high-temperature heating furnace and heated to 100℃, 150℃, and 200℃ respectively. After holding at these temperatures for 120 minutes, the specimens were allowed to cool to room temperature in the furnace and then left to stand for 48 hours.

[0040] c. The artificially aged specimens were subjected to uniaxial tensile tests, and real-time monitoring was performed using AE detection technology and IRT detection technology to obtain the time-domain amplitude diagram of the AE signal and the temperature-time history diagram of different specimens.

[0041] As shown in Figure 1, acoustic emission sensors are installed at both ends of the specimen, and the fracture location of the specimen is located on the material body.

[0042] The stress-strain curves of the strain-aged specimens are shown in Figure 2. Figures 2(a), 2(b), and 2(c) show the stress-strain curves of different pre-strained specimens at 100℃, 150℃, and 200℃, respectively. Compared with the un-strain-aged specimens, the mechanical properties change significantly. Although the fracture location of the welded joint specimen is located in the base material, the AE signal obtained by AE detection still includes several source mechanisms such as plastic deformation of the weld matrix, fracture of the weld matrix, and fracture of weld inclusions. Simply using the AE characteristic parameters cannot well distinguish the source mechanisms such as dislocation movement and slip within the material caused by strain aging. Therefore, we consider analyzing different strain-aged welded joint specimens from the perspective of waveform.

[0043] d. Perform a Fast Fourier Transform (FFT) on the AE time-domain signal acquired during the elastoplastic stage, and after post-processing, obtain the local energy ratio (PP) of a specific frequency band in the AE spectrum, and define the AE damage parameter DP.

[0044]

[0045] Wherein, Peak frequency is the peak frequency of the AE signal.

[0046] Typical AE time-domain signals are selected from the acoustic emission detection system, and the time-domain waveform signals are converted into single-column amplitude data. For each signal, a fast Fourier transform is performed to obtain the spectrum. Six frequency bands are defined, and the local energy ratio PP of a specific frequency band is calculated.

[0047] The calculation process for PP and DP values ​​is shown in Figure 3. The detection result of the AE technology for the uniaxial tensile process is the original time-domain waveform. When playing back the waveform, the characteristic signal waveform (AE signal time-domain amplitude diagram) is selected as shown in Figure 3a. The waveform is exported as a data file of single-column amplitude as shown in Figure 3b. The sampling time corresponding to each amplitude is determined by the sampling frequency. Then, the time-domain amplitude signal is subjected to a fast Fourier transform as shown in Figure 3c. The frequency range of 0kHz to 1200kHz is divided into 6 frequency intervals. The local energy percentage of the interval where the peak frequency is located is calculated. This value is substituted into the aforementioned formula to calculate the AE damage parameter DP as shown in Figure 3d.

[0048] In this invention, the sampling frequency is 3MHz. Calculate the local energy percentage PP of the AE signal in the 250kHz–500kHz range.

[0049]

[0050] in, It is a function obtained by performing a Fast Fourier Transform on the AE signal.

[0051] After obtaining the local energy value PP of a specific frequency band of the AE spectrum in the elastoplastic stage, it is substituted into the formula of the AE damage parameter DP to calculate the AE damage parameter DP.

[0052] e. Based on the temperature-time history diagram, calculate the temperature change rate ΔW during the strain hardening stage of the tensile process. The formula for calculating ΔW is:

[0053]

[0054] Where ΔW represents the rate of temperature change during the strain hardening stage, ΔT is the amount of temperature change, and Δt is the time increment.

[0055] Figure 4 shows the temperature change during uniaxial tensile testing. The three figures are the temperature-time curves of specimens after different strain aging treatments, measured by IRT, to calculate the temperature rise rate. Figure 4a is the temperature-time curve of the S10-100 specimen, Figure 4b is the temperature-time curve of the S10-150 specimen, and Figure 4c is the temperature-time curve of the S10-200 specimen.

[0056] f. Define the strain-aging damage index DI.

[0057] DI = Pre strain Aging temperature

[0058] Among them, Pre strain The term "pre-strain" refers to the pre-strain condition, and "aging temperature" refers to the aging temperature.

[0059] This discussion focuses on pre-strain and aging temperature. If aging time is considered, simply replace the temperature. With the same pre-strain, a higher aging temperature results in a stronger strain aging; with a constant aging temperature, a larger pre-strain results in a stronger strain aging.

[0060] g. Establish a neural network model, train the feedforward network using the backpropagation algorithm, and use the local energy ratio PP of a specific frequency band of the post-processed AE spectrum, the AE damage parameter DP, and the temperature change rate ΔW during the strain hardening stage as the input layer of the network, and the strain aging damage index DI as the output layer.

[0061] As shown in Figure 5, the neural network model of the present invention uses the backpropagation algorithm to train the feedforward network and uses a multi-layer network including an input layer, a hidden layer, and an output layer, where the output of one layer becomes the input of the next layer.

[0062] During the training of the feedforward network, the variables are normalized so that all variables are in the range of 0 to 1.

[0063] The backpropagation algorithm continuously reduces the loss function value through iterative calculations throughout the network training process, utilizing the gradient descent solution method.

[0064] After the fully connected neural network is trained, the network model will be integrated into the testing system, and its evaluation results will be incorporated into the security management system.

[0065] h. Obtain the test specimen and monitor the tensile testing process of the test specimen in real time using AE detection technology and IRT detection technology. Substitute the local energy ratio PP, AE damage parameter DP, and temperature rise rate ΔW of the strain hardening stage in a specific frequency band of the AE spectrum into the established neural network model to obtain the predicted strain aging damage index DI.

[0066] In safety evaluation, the tensile testing process of the sample is monitored in real time using two detection methods. By inputting the local energy of AE, AE damage parameters, and temperature rise rate into the system, the strain aging damage index can be obtained, which quantifies the strain aging degree of the welded steel structure.

[0067] This invention provides a testing method for assessing the degree of strain aging in welded steel structures. It employs both AE (Adaptive Aesthetic) and IRT (Integrated Thermal Reduction) testing technologies. IRT technology directly reflects the thermodynamics of metal plastic flow and strain localization processes, while AE technology offers ultra-high temporal resolution in characterizing the rapid release of elastic waves. The two technologies can compensate for each other when reflecting mechanical energy dissipation.

[0068] This invention involves tensile testing of welded joint specimens with different pre-strain and aging treatments, and real-time monitoring using AE (Advanced Electrode Spectrometry) and IRT (Intensity Reduction Test) technologies. This yields time-domain amplitude maps of the AE signals and temperature-time history maps, allowing for the determination of the input and output layers and the establishment of a neural network model. For each specimen under test, AE and IRT tests are performed, and the corresponding parameters are input into the established neural network model to evaluate the degree of strain aging. This method is not limited by the type of welding defect, sampling location, or number of samples. Furthermore, the two testing technologies compensate for each other in terms of mechanical energy dissipation, improving the accuracy of strain aging degree evaluation.

Claims

1. A method for evaluating the degree of strain aging in welded steel structures, characterized in that, Includes the following steps: a. First, two steel plates from the same batch are butted together by hand and cut into several specimens of the same shape, with the weld located in the middle of the specimen; b. The specimens are pre-stretched to achieve different levels of pre-strain, and the pre-stretched specimens are artificially aged at different temperatures. c. Perform uniaxial tensile tests on the artificially aged specimens and monitor them in real time using AE (Augmented Electrode) and IRT (Intensity Reduction) detection technologies to obtain the time-domain amplitude diagram and temperature-time history diagram of the AE signal; d. Perform Fast Fourier Transform (FFT) on the AE time-domain signal acquired during the elastoplastic stage, and after post-processing, obtain the local energy ratio (PP) of the frequency band containing the peak frequency in the AE spectrum, and define the AE damage parameter DP. Where, Peak frequency is the peak frequency of the AE signal; e. Calculate the temperature change rate ΔW during the strain hardening stage of the tensile process based on the temperature-time history diagram; f. Define the strain aging damage index DI. Among them, Pre strain g. Establish a neural network model, train the feedforward network using the backpropagation algorithm, and use the local energy ratio PP, AE damage parameter DP, and temperature change rate ΔW in the frequency band where the peak frequency of the post-processed AE spectrum is located as the input layer of the network, and the strain aging damage index DI as the output layer; h. Obtain the test specimen, and monitor the tensile test process of the test specimen in real time using AE detection technology and IRT detection technology. Substitute the local energy ratio PP, AE damage parameter DP, and temperature rise rate ΔW in the frequency band where the peak frequency of the AE spectrum is located into the established neural network model to obtain the predicted strain aging damage index DI.

2. The method for evaluating the strain aging degree of welded steel structures according to claim 1, characterized in that, In step d, typical AE time-domain signals are selected from the AE detection system, and the time-domain waveform signals are converted into single-column amplitude data. For each signal, a fast Fourier transform is performed to obtain a spectrum diagram. Six frequency band intervals are defined, and the local energy ratio PP of the frequency band interval where the peak frequency is located is calculated.

3. The method for evaluating the strain aging degree of welded steel structures according to claim 2, characterized in that, Find the local energy percentage PP of the AE signal in the 250kHz to 500kHz range. ;in, It is a function obtained by performing a Fast Fourier Transform on the AE signal.

4. The method for evaluating the strain aging degree of welded steel structures according to claim 1, characterized in that, During the training of the feedforward network, the variables are normalized so that all variables are in the range of 0 to 1.

5. The method for evaluating the strain aging degree of welded steel structures according to claim 1, characterized in that, In step b, the specimen is uniaxially stretched at room temperature to exceed the material yield point to achieve different levels of pre-strain, namely 6%, 10%, and 14%.

6. The method for evaluating the strain aging degree of welded steel structures according to claim 1, characterized in that, In step b, the pre-stretched specimen is placed in a high-temperature heating furnace and heated to 100°C, 150°C and 200°C respectively. After holding at these temperatures for 120 minutes, the specimen is cooled to room temperature in the furnace and left to stand for 48 hours.

Images