Method for estimating yield strength of ferromagnetic material based on electromagnetic ultrasound
Technical Field
The invention relates to the technical field of electromagnetic nondestructive testing, in particular to a method for estimating yield strength of a ferromagnetic material based on electromagnetic ultrasound.
Background
Ferromagnetic materials have good adaptability in the aspect of mechanical properties, so that the ferromagnetic materials are widely applied as key materials in the aspects of railways, transportation, energy sources, buildings, spaceflight, military, chemical engineering and the like.
However, a wide variety of ferromagnetic materials have very different mechanical physical and chemical properties due to different processing techniques, different carbon contents, metallurgical structures and different doped alloy component materials. In addition, the ferromagnetic material has a great difference in its behavior under different application environments, such as load, fatigue, corrosion, and usage temperature. Therefore, it is necessary to acquire material microstructure information by detecting the magnetic characteristics thereof, thereby determining the load, fatigue, and corrosion states thereof and acquiring material life information.
In addition, the ferromagnetic material can form defects under the long-term service condition, so that damage accidents are caused, and casualties and large economic loss are caused. The formation of defects includes incubation, initial crack initiation, and extension. Ferromagnetic materials exhibit microstructural changes that are usually due to various microscopic stress concentrations, etc., early in the damage, i.e., during the incubation period. Factors that cause defects in ferromagnetic materials during use include: local plastic deformation is finally caused by local stress and stress concentration caused by overlarge bearing, and defects are generated; fatigue damage from long-term use; the temperature change causes the metal to expand with heat and contract with cold, so that temperature stress accumulated in the material is formed; internal residual stress distribution caused by welding and other processing technologies; heat treatment parameters such as hardness, depth of carburized layer, austenite and martensite content, grain size, etc. Therefore, it is necessary to monitor the distribution of residual stresses and mechanical properties inside the material, so as to prevent local plastic deformation of the material and the generation of defects.
The magnetostrictive property of ferromagnetic materials is influenced by the microstructure of the material, mechanical strength, hardness, fatigue damage state, corrosion on the surface of the material, internal residual stress distribution, and changes in parameters of an applied magnetic field, temperature, etc. Therefore, the electromagnetic ultrasonic signal based on the magnetostrictive effect can reflect the change of the microstructure of the material such as fatigue damage, deformation size, defects and the like, and can be used for estimating the mechanical performance parameters (such as hardness, yield strength, tensile strength, elongation and the like) of the material.
At present, the method of nondestructive testing by using electromagnetic ultrasound is mainly used for detecting the defects borne by the material and positioning the defects, and meanwhile, the application of electromagnetic ultrasound in the aspect of detecting the wall thickness of the tube is widely researched. A Huazhong university of science and technology team led by Kangyihua teaching and professor of the Wuxin army has an outstanding research result in the aspect of researching the thickness measurement of the tube wall of the EMAT steel tube. They optimized the designed static bias magnetic field, and analyzed the method of replacing permanent magnet with electromagnet, and verified its superiority. The relation trend of the electromagnetic ultrasonic magnetic field intensity and the receiving end ultrasonic signal amplitude of the steel pipe defect test is analyzed by the Zhuxiu red professor of the Chinese mine big project team through experiments. However, the detection of mechanical properties of materials by electromagnetic ultrasound using the magnetostrictive effect has not been studied so far. Mechanical properties are a set of common indexes of metal materials, and are important material performance indexes used in the design of mechanical products. The quality of the use performance of the metal material determines the use range and the service life of the metal material, and the mechanical characteristics of the material are detected in an off-line stretching mode at present, so that the material is damaged, the reuse of the material is influenced, and the resource waste is caused.
Disclosure of Invention
The invention aims to solve the technical problem of providing an estimation method for the yield strength of a ferromagnetic material based on electromagnetic ultrasound aiming at the defect of detection of the mechanical property (yield strength) of the material in the prior art mentioned in the background art.
The invention adopts the following technical scheme for solving the technical problems:
the method for estimating the yield strength of the ferromagnetic material based on electromagnetic ultrasound comprises the following steps:
step 1), taking N ferromagnetic plates with known material yield strength, wherein N is a natural number greater than 0, and for each ferromagnetic plate:
step 1.1), an electromagnetic ultrasonic transmitting end and an electromagnetic ultrasonic receiving end are respectively arranged on a ferromagnetic plate, a first bias magnetic field is applied to the electromagnetic ultrasonic transmitting end, and a second bias magnetic field is applied to the electromagnetic ultrasonic receiving end, wherein the first bias magnetic field is formed by adopting an adjustable direct current electromagnet, the second bias magnetic field is formed by adopting excitation of a permanent magnet, the electromagnetic ultrasonic transmitting end transmits an electromagnetic ultrasonic signal, and the electromagnetic ultrasonic receiving end acquires the electromagnetic ultrasonic signal transmitted by the electromagnetic ultrasonic transmitting end;
step 1.2), gradually increasing the intensity of the first bias magnetic field from a preset first magnetic field intensity to a preset second magnetic field intensity according to a preset magnetic field intensity step length, reading the amplitude of an electromagnetic ultrasonic signal of an electromagnetic ultrasonic receiving end under each magnetic field intensity, fitting a magnetostrictive curve of the magnetic field intensity of the first bias magnetic field corresponding to the amplitude of the electromagnetic ultrasonic signal of the electromagnetic ultrasonic receiving end, extracting the amplitude of the electromagnetic ultrasonic signal of the electromagnetic ultrasonic receiving end corresponding to the magnetic field intensity of the first bias magnetic field in the magnetostrictive curve as a first characteristic value, the magnetic field intensity of the first bias magnetic field corresponding to a first valley point as a second characteristic value, the magnetic field intensity of the first bias magnetic field corresponding to a first peak point as a third characteristic value, and the magnetic field intensity of the first bias magnetic field corresponding to a second valley point as a fourth characteristic value;
step 2), establishing a neural network database according to the attribute data of the N ferromagnetic plates, wherein the attribute data of the ferromagnetic plates comprise the yield strength, the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of the material of the ferromagnetic plates;
step 3), dividing data in the neural network database into a training sample set and a testing sample set, and normalizing the data in the training sample set and the testing sample set to be between [ -1,1 ];
step 4), establishing a BP neural network model, taking the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of each ferromagnetic plate in the training sample set as input, taking the material yield strength of each ferromagnetic plate in the training sample set as output, training the BP neural network model, and finishing training when the training error is smaller than a preset first error threshold value to obtain a trained BP neural network model;
step 5), inputting the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of each ferromagnetic plate in the test sample set into a trained BP neural network model to obtain a material yield strength estimation value of each ferromagnetic plate in the test sample set;
step 6), calculating the relative error between the material yield strength and the material yield strength estimated value of each ferromagnetic plate in the test sample set, and if the relative error between the material yield strength and the material yield strength estimated value of the ferromagnetic plate is smaller than a preset second error threshold, determining that the ferromagnetic plate is qualified, and determining that the ferromagnetic plate is unqualified if the relative error is not smaller than the preset second error threshold;
step 7), calculating the qualified rate of the ferromagnetic plates in the test sample set, namely dividing the number of the qualified ferromagnetic plates by the number of the ferromagnetic plates in the test sample set, and skipping to the step 1 if the qualified rate of the ferromagnetic plates in the test sample set is less than or equal to a preset qualified rate threshold value;
and 8), inputting the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of the ferromagnetic plate needing to be subjected to the material yield strength into the trained BP neural network model to obtain a material yield strength estimation value of the ferromagnetic plate needing to be subjected to the material yield strength.
As a further optimization scheme of the method for estimating the yield strength of the ferromagnetic material based on electromagnetic ultrasound, the preset first magnetic field strength is 0 tesla.
As a further optimization scheme of the method for estimating the yield strength of the ferromagnetic material based on the electromagnetic ultrasound, the frequency of an excitation signal of the ultrasonic wave transmitting end is 200KHz, and the number of pulses is 8.
As a further optimization scheme of the method for estimating the yield strength of the ferromagnetic material based on electromagnetic ultrasound, the preset first error threshold is 0.01.
As a further optimization scheme of the method for estimating the yield strength of the ferromagnetic material based on electromagnetic ultrasound, the preset second error threshold is 10%.
As a further optimization scheme of the method for estimating the yield strength of the ferromagnetic material based on electromagnetic ultrasound, the preset qualification rate threshold is 80%.
As a further optimization scheme of the method for estimating the yield strength of the ferromagnetic material based on electromagnetic ultrasound, the BP neural network model comprises an input layer, two hidden layers and an output layer, wherein the input layer comprises 5 nodes, one hidden layer comprises 9 nodes, the other hidden layer comprises 3 nodes, and the output layer comprises 1 node.
As a further optimization scheme of the method for estimating the yield strength of the ferromagnetic material based on electromagnetic ultrasound, the ratio of the number of samples in the training sample set and the number of samples in the testing sample set in the step 3) is 4: 1.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects:
1. by establishing a BP neural network model, the yield strength of the ferromagnetic material can be quantitatively estimated;
2. the method can achieve higher detection yield, and changes the detection mode of destructive online stretching on the mechanical properties of the material, such as yield strength.
Drawings
FIG. 1 is a diagram of an ultrasonic signal received by a receiving end of an electromagnetic ultrasonic inspection system according to the present invention;
FIGS. 2 (a) and (b) are respectively a graph of magnetostrictive characteristic and electromagnetic ultrasonic signal of the material of the present invention;
FIG. 3 is a diagram of the variation of the amplitude of the electromagnetic ultrasonic receiving signal with the applied magnetic field according to the present invention;
fig. 4 is a schematic diagram of relative errors between the yield strengths of the ferromagnetic plate materials and the estimated values of the yield strengths of the ferromagnetic plate materials in the test sample set according to the present invention.
Detailed Description
The technical scheme of the invention is further explained in detail by combining the attached drawings:
the present invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. In the drawings, components are exaggerated for clarity.
Ferromagnetic substances have a crystal-like structure. Between adjacent atoms, magnetic element moments are generated due to electron spin, and there is an interaction force between the magnetic element moments, which drives the adjacent magnetic element moments to be arranged in parallel in the same direction, forming a magnetic domain. In the absence of external magnetic field, the magnetic domains are balanced, and the total magnetization of the material is equal to zero. When an external magnetic field acts, the magnetic domain rotates, so that the length or volume of the material slightly changes, and the phenomenon is called magnetostrictive effect. Different ferromagnetic materials have different magnetostriction characteristics, the magnetostriction characteristics are influenced by the microstructure of the material, an external magnetic field, material stress, a heat treatment state and the like, and an electromagnetic ultrasonic signal based on the magnetostriction effect is influenced by the magnetostriction characteristics of the ferromagnetic materials, so that the magnetostriction characteristics of the materials can be reflected according to the electromagnetic ultrasonic signal, and the relationship between the two characteristics is shown in fig. 2.
The invention discloses a method for estimating yield strength of a ferromagnetic material based on electromagnetic ultrasound, which comprises the following steps:
step 1), taking 50 ferromagnetic plates with known material yield strength, and for each ferromagnetic plate:
step 1.1), as shown in fig. 1, respectively arranging an electromagnetic ultrasonic transmitting end and an electromagnetic ultrasonic receiving end on a ferromagnetic plate, applying a first bias magnetic field to the electromagnetic ultrasonic transmitting end, and applying a second bias magnetic field to the electromagnetic ultrasonic receiving end, wherein the first bias magnetic field is formed by adopting an adjustable direct current electromagnet, the second bias magnetic field is formed by adopting permanent magnet excitation, the electromagnetic ultrasonic transmitting end transmits an electromagnetic ultrasonic signal, and the electromagnetic ultrasonic receiving end acquires the electromagnetic ultrasonic signal transmitted by the electromagnetic ultrasonic transmitting end; the frequency of an excitation signal of the ultrasonic transmitting end is 200KHz, and the number of pulses is 8;
step 1.2), gradually increasing the intensity of the first bias magnetic field from 0 Tesla to a preset second magnetic field intensity according to a preset magnetic field intensity step length, reading the amplitude of an electromagnetic ultrasonic signal of an electromagnetic ultrasonic receiving end under each magnetic field intensity, and fitting a magnetostrictive curve of the magnetic field intensity of the first bias magnetic field corresponding to the amplitude of the electromagnetic ultrasonic signal of the electromagnetic ultrasonic receiving end, as shown in FIG. 3;
extracting the amplitude of an electromagnetic ultrasonic signal of a corresponding electromagnetic ultrasonic receiving end when the intensity of a first bias magnetic field in a magnetostrictive curve is 0 as a first characteristic value, wherein the magnetic field strength of the first bias magnetic field corresponding to a first valley point is a second characteristic value, the magnetic field strength of the first bias magnetic field corresponding to a first peak point is a third characteristic value, and the magnetic field strength of the first bias magnetic field corresponding to a second valley point is a fourth characteristic value;
step 2), establishing a neural network database according to the attribute data of the N ferromagnetic plates, wherein the attribute data of the ferromagnetic plates comprise the yield strength, the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of the material of the ferromagnetic plates;
step 3), dividing data in the neural network database into a training sample set and a test sample set, specifically, dividing attribute data of 40 ferromagnetic plates into the training sample set, dividing attribute data of the remaining 10 ferromagnetic plates into the test sample set, and normalizing the data in the training sample set and the test sample set to be between [ -1,1 ];
step 4), establishing a BP neural network model, wherein the BP neural network model comprises an input layer, two hidden layers and an output layer, the input layer comprises 5 nodes, one hidden layer comprises 9 nodes, the other hidden layer comprises 3 nodes, and the output layer comprises 1 node;
taking the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of each ferromagnetic plate in the training sample set as input, taking the material yield strength of each ferromagnetic plate in the training sample set as output, training the BP neural network model, and when the training error is less than 0.01, finishing the training to obtain the trained BP neural network model;
step 5), inputting the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of each ferromagnetic plate in the test sample set into a trained BP neural network model to obtain a material yield strength estimation value of each ferromagnetic plate in the test sample set;
step 6), calculating the relative error between the material yield strength and the material yield strength estimated value of each ferromagnetic plate in the test sample set, and if the relative error between the material yield strength and the material yield strength estimated value of the ferromagnetic plate is less than 10%, determining that the ferromagnetic plate is qualified, and determining that the ferromagnetic plate is unqualified if the relative error is not more than 10%; FIG. 4 is a graph illustrating the relative error between the material yield strength and the estimated material yield strength of each ferromagnetic plate in a set of test samples;
step 7), calculating the qualified rate of the ferromagnetic plates in the test sample set, namely dividing the qualified number of the ferromagnetic plates by the number of the ferromagnetic plates in the test sample set, and skipping to the step 1 if the qualified rate of the ferromagnetic plates in the test sample set is less than or equal to 80%;
and 8), inputting the thickness, the first characteristic value, the second characteristic value, the third characteristic value and the fourth characteristic value of the ferromagnetic plate needing to be subjected to the material yield strength into the trained BP neural network model to obtain a material yield strength estimation value of the ferromagnetic plate needing to be subjected to the material yield strength.
The invention aims to provide a method for estimating the yield strength of a ferromagnetic material by using electromagnetic ultrasound based on a magnetostrictive effect, and a network model is established by using a neural network, so that the nondestructive quantitative detection of the mechanical property of the yield strength of the material is realized.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The above-mentioned embodiments, objects, technical solutions and advantages of the present invention are further described in detail, it should be understood that the above-mentioned embodiments are only illustrative of the present invention and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.