A construction management method based on extremely weak magnetic monitoring

Abstract

The present application discloses a construction management method based on extremely weak magnetic monitoring, which relates to the field of material detection, including: obtaining extremely weak magnetic materials to be detected at the construction site; wherein the extremely weak magnetic materials include non-magnetic stainless steel and aluminum alloy; using a nuclear magnetic resonance spectrometer to perform sampling residual magnetism detection on the obtained extremely weak magnetic materials to be detected; entering the detection data into the material management system; wherein the nuclear magnetic resonance spectrometer includes a Z-direction gradient coil, and a multi-channel room temperature shim coil in the X-direction and the Y-direction; according to the detection results, the extremely weak magnetic materials that have passed the detection are marked as available for construction use; otherwise, they are marked as unavailable. In view of the low precision of quantitative detection of extremely weak magnetic materials in the prior art, the present application uses a nuclear magnetic resonance spectrometer to sample and detect the residual magnetism distribution on the surface and inside of the material by setting a three-dimensional gradient echo and spin echo pulse sequence, thereby improving the precision of quantitative detection.

Description

Technical Field

[0001] The present application relates to the field of material detection, and in particular to a construction management method based on extremely weak magnetic monitoring. Background Art

[0002] With the rapid development of high-tech fields such as precision manufacturing, aerospace and military industry, higher and higher requirements are placed on the extremely weak magnetic properties of materials and tools. Extremely weak magnetic materials are widely used in precision instruments, aerospace equipment, biomedical devices and other occasions that are highly sensitive to magnetic interference. For example, when assembling precision payloads in satellite cabins, extremely weak magnetic tools are required to operate to avoid residual magnetism from interfering with sensitive devices; in the manufacturing process of nuclear magnetic resonance imaging equipment, the materials and tools used must meet the standards of extremely low residual magnetism to ensure imaging quality. Therefore, it is crucial to perform high-precision and comprehensive residual magnetism detection on extremely weak magnetic materials and tools.

[0003] At present, the residual magnetism detection of materials and tools mainly adopts conventional methods such as fluxgate method, Hall probe method, magnetic force microscope, etc. Although these methods can realize the residual magnetism measurement of local areas on the surface of materials, there are many blind areas in the detection, and it is difficult to obtain the distribution information of the residual magnetism inside the material, resulting in incomplete evaluation of extremely weak magnetic properties. At the same time, the existing detection methods lack quantitative analysis methods, and mainly rely on manual experience to make qualitative judgments on the qualification of materials, which are easily affected by subjective factors, resulting in inconsistent material screening standards. In addition, conventional detection usually requires pretreatment of materials such as demagnetization, and the detection process is cumbersome and time-consuming, which is not conducive to engineering applications. These problems restrict the large-scale use and industrial development of extremely weak magnetic materials.

[0004] Taking the fasteners used in satellite precision assembly as an example, aerospace engineering has strict requirements on the residual magnetism level of fasteners such as screws and nuts, usually requiring the surface residual magnetism peak value not to exceed 1nT and the average residual magnetism intensity of the body to be less than 0.5nT. Traditional detection methods are difficult to achieve a comprehensive characterization of the residual magnetism on the thread surface and inside the screw, and lack clear quantitative indicators. It is impossible to ensure that the selected fasteners fully meet the use requirements, and there is a certain application risk. Therefore, a new method for detecting the residual magnetism of extremely weak magnetic materials and tools with high precision and strong applicability is urgently needed. Summary of the invention

[0005] In response to the problem of low precision in quantitative detection of extremely weak magnetic materials in the prior art, the present application provides a construction management method based on extremely weak magnetic monitoring, which uses a nuclear magnetic resonance spectrometer to set three-dimensional gradient echo and spin echo pulse sequences to sample and detect the surface and internal residual magnetic distribution of the material, thereby improving the quantitative detection accuracy.

[0006] The purpose of this application is achieved through the following technical solutions.

[0007] The present application provides a construction management method based on extremely weak magnetic monitoring, comprising: obtaining extremely weak magnetic materials to be tested at a construction site; wherein the extremely weak magnetic materials include non-magnetic stainless steel and aluminum alloy; using a nuclear magnetic resonance spectrometer to perform sampling residual magnetism detection on the obtained extremely weak magnetic materials to be tested; entering the detection data into a material management system; wherein the nuclear magnetic resonance spectrometer includes a Z-direction gradient coil, and multi-channel room temperature uniform field coils in the X-direction and the Y-direction; according to the detection results, marking the extremely weak magnetic materials that have passed the detection as being in an available state for use in construction; otherwise, marking them as being in an unavailable state.

[0008] Among them, extremely weak magnetic materials: refers to materials with very low residual magnetization intensity, usually in the order of 0.1 to 10 mT, and also weak response to external magnetic field. Common extremely weak magnetic materials include non-magnetic stainless steel, aluminum alloy, aluminum, titanium, etc. Such materials are widely used in precision instruments, aerospace, nuclear physics and other fields, and have strict requirements on the weak magnetic properties of materials. Z-direction gradient coil: In nuclear magnetic resonance spectrometers, coils used to generate magnetic field gradients in the Z direction (i.e., the direction of static magnetic field). By changing the current of the Z-direction gradient coil, the magnetic field intensity in the Z direction can be modulated to change linearly, which is used for spatial encoding and slice selection. Multi-channel room temperature shim coils in the X and Y directions: In nuclear magnetic resonance spectrometers, coil groups used to generate uniform magnetic fields in the X and Y directions (i.e., perpendicular to the static magnetic field direction). Multi-channel means that the coil is divided into multiple independent channels along the circumferential direction, and the current of different channels can be controlled separately to compensate for the inhomogeneity of the magnetic field on the sample surface. Room temperature shim means that the coil works at room temperature, without superconducting refrigeration, which is convenient for rapid placement and removal of samples.

[0009] Furthermore, a nuclear magnetic resonance spectrometer is used to perform a sampling residual magnetism test on the acquired extremely weak magnetic material to be tested, including: S1, placing the material to be tested at the center position in the detection chamber of the nuclear magnetic resonance spectrometer, and making the magnetization direction of the material to be tested parallel to the direction of the external static magnetic field; placing the material to be tested directly at the center position in the detection chamber of the nuclear magnetic resonance spectrometer without using the sample tube wall to avoid interference that may be introduced by the sample tube wall; directly performing three-dimensional imaging and analysis on the material to be tested, without the need for the signal on the surface of the sample tube wall to reflect the material itself, which is more direct and accurate. S2, setting a three-dimensional gradient echo pulse sequence and a spin echo pulse sequence on the nuclear magnetic resonance spectrometer; adding a spin echo pulse sequence, using its characteristic of compensating for magnetic field inhomogeneity, can effectively eliminate artifacts and distortions caused by magnetic field inhomogeneity inside and on the surface of the sample, and improve the quality of the residual magnetism distribution image. S3, successively executing the set three-dimensional gradient echo pulse sequence and spin echo pulse sequence, respectively collecting nuclear magnetic resonance signals, and obtaining three-dimensional spatial signal data and one-dimensional spin echo signal data of the material to be detected; S4, performing three-dimensional Fourier transform on the obtained three-dimensional spatial signal data of the material to be detected, and obtaining three-dimensional spatial frequency domain data of the material to be detected; according to the geometric size of the material to be detected, extracting two-dimensional frequency domain data corresponding to the surface of the material to be detected from the three-dimensional spatial frequency domain data; performing inverse Fourier transform on the extracted two-dimensional frequency domain data, and reconstructing a two-dimensional residual magnetic distribution image of the surface of the material to be detected; The one-dimensional spin echo signal data is subjected to a one-dimensional Fourier transform to obtain the one-dimensional frequency domain data of the material to be detected, and the volume average residual magnetic intensity of the material to be detected is calculated based on the one-dimensional frequency domain data; S5, respectively calculates the uniformity parameters of the reconstructed two-dimensional residual magnetic distribution image of the surface of the material to be detected and the volume average residual magnetic intensity of the material to be detected; based on the uniformity of the surface residual magnetic distribution and the volume average residual magnetic intensity, comprehensively judge whether the material to be detected meets the requirements of the extremely weak magnetic material; if the uniformity of the surface residual magnetic distribution is higher than the first preset threshold and the volume average residual magnetic intensity is lower than the second preset threshold, it is judged as a qualified extremely weak magnetic material, otherwise it is unqualified. On the basis of the uniformity of the surface residual magnetic distribution, the volume average residual magnetic intensity index is added, which can more comprehensively evaluate the overall residual magnetic level of the material and avoid focusing only on the one-sidedness of the surface. By setting two thresholds for surface uniformity and volume average intensity for comprehensive judgment, the residual magnetic characteristics requirements of the surface and the interior of the material can be more reasonably taken into account.

[0010] Among them, the three-dimensional gradient echo pulse sequence: a commonly used nuclear magnetic resonance imaging pulse sequence, which encodes the three-dimensional spatial position of the sample by applying magnetic field gradients in the three directions of X, Y, and Z to obtain the three-dimensional structural information of the sample. The echo in the sequence refers to a series of attenuated signal echoes generated by applying an anti-phase pulse to refocus the magnetization vector after one excitation, which can be used to improve the signal-to-noise ratio and shorten the imaging time. Spin echo pulse sequence: a commonly used nuclear magnetic resonance wave pulse sequence, mainly used to detect the chemical shift information of atomic nuclei under different chemical environments in the sample. In the sequence, a 90° radio frequency pulse is first applied to excite the nuclear spin, and then a 180° radio frequency pulse is applied to invert the nuclear spin to generate a spin echo signal. By collecting the signal intensity at different echo times, the nuclear magnetic resonance spectrum of the sample can be obtained. Three-dimensional Fourier transform: a mathematical transformation method that converts signals in the time domain or space domain to the frequency domain. In nuclear magnetic resonance imaging, by performing a three-dimensional Fourier transform on the collected three-dimensional spatial signal data, the distribution of the sample in the three-dimensional spatial frequency domain can be obtained, reflecting the frequency characteristics of the internal structure of the sample. Three-dimensional spatial frequency domain data: data representation of the sample in the three-dimensional spatial frequency domain. Through Fourier transform, the three-dimensional spatial signal data of the sample can be converted into three-dimensional spatial frequency domain data. In the frequency domain, each dimension of the data corresponds to a frequency component in a spatial direction. The higher the frequency, the faster the spatial change. The characteristic information of the surface and interior of the sample can be easily extracted through frequency domain data. Uniformity parameter: an indicator used to quantitatively evaluate the uniformity of the residual magnetic distribution on the surface of the material. Common uniformity parameters include standard deviation, coefficient of variation, root mean square, etc., which reflect the degree of discreteness of the residual magnetic intensity in spatial distribution. The smaller the uniformity parameter, the more uniform the residual magnetic distribution on the surface of the material, and the smaller the difference between the regions. Uniformity is one of the important parameters for evaluating the extremely weak magnetic properties of materials.

[0011] Further, S1, the material to be tested is directly placed in the center of the detection chamber of the nuclear magnetic resonance spectrometer, and the magnetization direction of the material to be tested is parallel to the direction of the external static magnetic field, including: S11, at least three magnetization marks with different directions are set on the outer surface of the material to be tested, each magnetization mark is made of a strong magnetic film material with a known magnetization direction and magnetization intensity, and the magnetization directions of any two magnetization marks are not parallel; the number requirement of the magnetization marks and the limitation that the magnetization directions are not parallel to each other are increased, which can avoid the direction calculation error caused by the coplanarity of the magnetization marks and improve the accuracy of the magnetization direction measurement. S12, the nuclear magnetic resonance spectrometer is used to image each magnetization mark on the material to be tested separately, and the spatial position, magnetization direction and magnetization intensity of each magnetization mark are obtained; the magnetization intensity of the magnetization mark is introduced as a calculation parameter, and the magnetization direction of the whole material can be more accurately calculated by intensity weighted averaging. S13, based on the spatial position, magnetization direction and magnetization intensity of each magnetization mark, a three-dimensional magnetization direction distribution model of the material to be tested is constructed, and the magnetization direction vector of the material to be tested is calculated; the three-dimensional magnetization direction distribution model of the material is constructed, which can realize the vector representation and quantitative calculation of the material magnetization direction, and provide a mathematical basis for subsequent adjustments. S14, based on the magnetization direction vector of the material to be tested and the external static magnetic field direction vector, the angle between the two vectors is calculated to obtain the angle value of the magnetization direction deviating from the static magnetic field direction; the magnetic field gradient pulse is used to apply magnetic torque to realize the active adjustment of the material magnetization direction, which is more precise and controllable than mechanical adjustment, and can minimize the influence of human factors. S15, according to the magnetization direction deviation angle value, use the gradient coil of the nuclear magnetic resonance spectrometer to apply a magnetic field gradient pulse opposite to the deviation direction, apply a magnetic torque to the material to be tested, so that its magnetization direction gradually approaches the static magnetic field direction; S16, repeat steps S12 to S15 until the calculated magnetization direction deviation angle value is less than the preset angle threshold; S17, stop applying the magnetic field gradient pulse, and fix the material to be tested that meets the magnetization direction adjustment requirements at the center position in the detection cavity of the nuclear magnetic resonance spectrometer. The magnetization direction adjustment is converted into a closed-loop feedback control, and high-precision magnetization direction alignment can be adaptively achieved by real-time detection of the deviation angle and continuous application of the correction torque.

[0012] Among them, the magnetization marker is a strong magnetic film with known magnetization direction and magnetization intensity attached to the surface of the material to be detected. By imaging and measuring the magnetization marker, the magnetization direction distribution of the material to be detected in space can be determined. The magnetization marker acts as a spatial reference. By calculating the angle between the magnetization direction of the marker and the static magnetic field direction, the degree of deviation of the magnetization direction of the material to be detected can be quantified, and a magnetic field gradient pulse can be applied for adjustment. Generally, soft magnetic materials with high saturation magnetization intensity and small magnetic anisotropy are selected as magnetization markers, such as Permalloy, iron-based amorphous, etc.

[0013] Further, S2, setting a three-dimensional gradient echo pulse sequence and a spin echo pulse sequence on the nuclear magnetic resonance spectrometer, including: S21, setting a three-dimensional gradient echo pulse sequence, the three-dimensional gradient echo pulse sequence includes a first radio frequency pulse, a first X-direction variable gradient pulse, a first Y-direction variable gradient pulse, a first Z-direction variable gradient pulse and a first sampling window; S22, setting a spin echo pulse sequence, the spin echo pulse sequence includes a second radio frequency pulse, a second Z-direction variable gradient pulse, a third Z-direction variable gradient pulse, a first shear gradient pulse, a second shear gradient pulse and a second sampling window; S23, calculating the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the second sampling window according to the magnetic susceptibility anisotropy of the material to be detected and the required spatial resolution of the residual magnetization distribution. The maximum amplitude, amplitude gradient rate and duration of the pulse and the first Z-direction variable gradient pulse are used to optimize the three-dimensional spatial encoding effect; S24, according to the relaxation time constant T1 and the relaxation time constant T2 of the material to be detected, the flip angles of the first radio frequency pulse and the second radio frequency pulse are calculated, and the calculation results are used to set the flip angles of the two radio frequency pulses in S21 and S22 respectively, so as to reduce the signal attenuation caused by longitudinal and transverse relaxation while satisfying the nuclear spin excitation and inversion; at the same time, the time interval between the second radio frequency pulse and the first radio frequency pulse is calculated, and the calculation result is used to set the application time of the second radio frequency pulse in S22, so as to avoid the first sampling data signal from decaying to the noise level before the phase inversion due to insufficient relaxation time constant T2.

[0014] Among them, relaxation time constant: describes the dynamic parameters of the nuclear spin system returning to equilibrium after being disturbed by a radio frequency pulse. The relaxation time constant includes the longitudinal relaxation time constant T1 and the transverse relaxation time constant T2. T1 describes the time required for the longitudinal magnetization component (i.e. parallel to the static magnetic field direction) to recover from the non-equilibrium state to the equilibrium state; T2 describes the time required for the transverse magnetization component (i.e. perpendicular to the static magnetic field direction) to lose coherence due to the interaction between spins. In general, T1 is greater than T2. ​​The relaxation time constant depends on the chemical composition, microstructure and environment of the material. The dynamic properties of the material can be characterized by measuring the relaxation curve. In the pulse sequence design, the time interval and sampling time of the radio frequency pulse need to be optimized according to the relaxation time constant of the material to be detected in order to obtain sufficient signal strength and signal-to-noise ratio. Flip angle: the angle at which the nuclear spin magnetization vector rotates relative to the static magnetic field direction after the radio frequency pulse acts. Common flip angles include 90° (i.e. π / 2) and 180° (i.e. π). A 90° pulse can rotate the longitudinal magnetization parallel to the static magnetic field in the equilibrium state to the vertical plane, forming a transverse magnetization; a 180° pulse can reverse the transverse magnetization by 180°, forming a spin echo. By controlling the power and duration of the RF pulse, the flip angle of the nuclear spin can be precisely controlled. In the pulse sequence design, the selection of the flip angle needs to comprehensively consider the number of nuclear spins, relaxation characteristics and required magnetization intensity, so as to minimize the relaxation loss while satisfying the excitation or reversal conditions. Nuclear spin excitation: Under the action of the RF pulse, the nuclear spin system absorbs the energy of the RF field, causing the nuclear spin to transition from a low energy level to a high energy level. Nuclear spin excitation is the basis of the nuclear magnetic resonance phenomenon. According to the quantum mechanical selection rule, effective nuclear spin excitation can only occur when the frequency of the RF field is equal to the Larmor precession frequency of the nuclear spin. After excitation, the nuclear spin magnetization vector will deviate from the static magnetic field direction and precess, forming a component of the macroscopic magnetization in the transverse plane. The degree of nuclear spin excitation depends on the power, frequency and duration of the RF pulse. In pulse sequence design, 90° RF pulses are usually used to excite the nuclear spin system and transfer the longitudinal equilibrium magnetization to the transverse plane for subsequent spatial encoding and signal acquisition.

[0015] Further, S21, a three-dimensional gradient echo pulse sequence is set, and the three-dimensional gradient echo pulse sequence includes a first radio frequency pulse, a first X-direction variable gradient pulse, a first Y-direction variable gradient pulse, a first Z-direction variable gradient pulse and a first sampling window, including: the flip angle of the first radio frequency pulse is set in the range of 30° to 90° according to the calculation result of S24; the output of the first radio frequency pulse is used to excite nuclear spins of different chemical shifts in the material to be detected, so as to generate a transverse magnetization component as an input signal of the first sampling window; the maximum amplitude, amplitude gradient rate and duration of the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the first Z-direction variable gradient pulse are set according to the calculation result of S23, and the output of the three XYZ variable gradient pulses is used to perform dual encoding of the spatial position and precession frequency of the nuclear spins generated by the first radio frequency pulse excitation, so as to form a nuclear magnetic resonance signal with spatial position information and frequency information as the input signal of the first sampling window; the first sampling window obtains first sampling data with different spatial resolutions and signal-to-noise ratios by controlling the number of sampling points and the sampling time.

[0016] Among them, transverse magnetization component: after the nuclear spin system is excited by the radio frequency pulse, the magnetization vector will deviate from the direction of the static magnetic field and form a component in the transverse plane perpendicular to the static magnetic field, which is called the transverse magnetization component. The transverse magnetization component precesses in the static magnetic field, and its precession frequency is equal to the Larmor frequency, which can induce a nuclear magnetic resonance signal through the sampling coil. The magnitude of the transverse magnetization component is proportional to the sine value of the flip angle of the radio frequency pulse, and decays exponentially with the relaxation time constant T2 over time. One of the purposes of pulse sequence design is to maximize the transverse magnetization component to obtain a higher signal intensity. Precession frequency: the frequency at which the nuclear spin magnetic moment precesses in an external static magnetic field, also known as the Larmor frequency. According to the Larmor equation, the precession frequency is proportional to the magnetic gyro ratio of the nucleus and the strength of the external static magnetic field. Atomic nuclei in different chemical environments have different shielding effects, and the local magnetic field strengths they feel are different, so they show different precession frequencies, which is the frequency basis of the chemical shift. Applying a magnetic field gradient in the pulse sequence can form a precession frequency gradient at different spatial positions of the sample to achieve spatial encoding. By collecting NMR signals with different precession frequencies and combining them with spatial position information, the frequency and position distribution of the sample can be obtained.

[0017] Further, S21, the output of the three variable gradient pulses XYZ is used to perform dual encoding of the spatial position and precession frequency of the nuclear spin generated by the first radio frequency pulse excitation, including: the first radio frequency pulse adopts the form of a Gaussian modulated soft pulse, and the duration and bandwidth of the soft pulse are set according to the number of different chemical shifts present in the material; the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the first Z-direction variable gradient pulse adopt the form of gradual amplitude change and increasing duration, and by controlling the rate of gradual amplitude change and the increment of duration, a magnetic field with increasing gradient is formed at different spatial positions, so that the precession frequency of the nuclear spin forms a mapping relationship with the corresponding spatial position; the first sampling window is set to be synchronized with the application of the first X-direction variable gradient pulse, and by controlling the starting time and duration of the sampling window, a time domain nuclear magnetic resonance signal corresponding to the spatial position and the precession frequency is obtained, and is converted into a spatial frequency domain signal through fast Fourier transform to perform spatial encoding on the first sampling data.

[0018] Among them, the soft pulse form of Gaussian modulation: a modulation method of radio frequency pulse, which makes the envelope of radio frequency pulse present the shape of Gaussian function (normal distribution). Compared with the ideal rectangular hard pulse, the Gaussian soft pulse has the characteristics of more concentrated spectrum and controllable pulse bandwidth. In samples with multiple chemical shifts, the use of Gaussian soft pulses can selectively excite nuclear spins in a specific frequency range and suppress the nuclear spin response of other frequencies, thereby obtaining the local signal of the target chemical shift and improving the spectral peak resolution. The duration and bandwidth of the Gaussian soft pulse can be optimized according to the number and distribution characteristics of the chemical shift of the sample. The number of chemical shifts: the number of types of atomic nuclei in different chemical environments in the sample. The different chemical environments of the atomic nuclei, such as the types of atoms bonded to them, bond lengths, bond angles, etc., will affect the local magnetic field felt by the atomic nuclei, produce different degrees of shielding effects, and manifest as chemical shift differences on the nuclear magnetic resonance spectrum. The number of chemical shifts depends on the molecular structure and composition complexity of the sample. The more chemical shifts there are, the more peaks appear on the nuclear magnetic resonance spectrum, and the wider the corresponding frequency band range. In pulse sequence design, it is necessary to reasonably set the modulation mode, duration and bandwidth of the RF pulse according to the number of chemical shifts, so as to achieve selective excitation and acquisition of different chemical shifts and improve the specificity and sensitivity of the target signal.

[0019] Further, S22, a spin echo pulse sequence is set, and the spin echo pulse sequence includes a second radio frequency pulse, a second Z-direction variable gradient pulse, a third Z-direction variable gradient pulse, a first shear gradient pulse, a second shear gradient pulse and a second sampling window, including: the effective flip angle of the second radio frequency pulse is set in the range of 175° to 185° according to the calculation result of S24, and the output of the second radio frequency pulse is used to simultaneously invert the phases of nuclear spins with different chemical shifts in the first sampling data to suppress the residual phase difference caused by insufficient inversion pulse bandwidth; the second Z-direction variable gradient pulse is applied before the second radio frequency pulse, and the output is the same as the first Z-direction variable gradient pulse; the third Z-direction variable gradient pulse is applied after the second radio frequency pulse; the application timing and amplitude change form of the second Z-direction variable gradient pulse and the third Z-direction variable gradient pulse are the same as those of the first Z-direction variable gradient pulse; the first shear gradient pulse and the second shear gradient pulse are applied before and after the second radio frequency pulse respectively, and the amplitude size and application time are calculated and set according to the distribution of magnetic field gradients between different cross sections of the material, and the two shear pulse gradient directions are opposite; the duration of the second sampling window is the same as that of the first sampling window.

[0020] Among them, the first shear gradient pulse: a magnetic field gradient pulse applied before the 180° radio frequency pulse in the spin echo pulse sequence. The function of the first shear gradient pulse is to generate a linearly changing magnetic field on different sections of the sample (i.e., the plane perpendicular to the gradient direction), so that the nuclear spin precession frequencies on these sections differ. This frequency difference will be reversed after the 180° pulse, so that under the action of the second shear gradient pulse, the nuclear spins in different sections will reunite to form a spin echo. The gradient directions of the two shear gradient pulses are opposite, and the amplitudes and durations are equal to achieve precise spin reunion. The parameter setting of the shear gradient pulse needs to be optimized according to the size, shape and internal magnetic field distribution inhomogeneity of the sample to obtain high-quality spin echo signals.

[0021] Furthermore, the second RF pulse is in the form of a wideband composite pulse, including: multiple sub-pulses, the frequency, phase and flip angle of each sub-pulse are set according to the nuclear spin resonance frequency range and chemical shift difference of the material, and a uniform inversion effect is obtained by changing the number and parameters of sub-pulses.

[0022] Further, S3, successively executes the set three-dimensional gradient echo pulse sequence and spin echo pulse sequence, respectively collects nuclear magnetic resonance signals, and obtains three-dimensional spatial signal data and one-dimensional spin echo signal data of the material to be detected, including: according to the set three-dimensional gradient echo pulse sequence, executes a three-dimensional gradient echo pulse sequence once, and collects first sampling data obtained in a first sampling window as the three-dimensional spatial signal data of the material to be detected; according to the set spin echo pulse sequence, after the three-dimensional gradient echo pulse sequence is executed, executes a spin echo pulse sequence once, and collects second sampling data obtained in a second sampling window as the one-dimensional spin echo signal data of the material to be detected; wherein the first sampling data reflects the distribution of nuclear spins at different spatial positions on the surface and inside of the material to be detected; and the second sampling data reflects the resonant frequency distribution of the overall nuclear spin of the material to be detected after spin echo.

[0023] Further, S4, calculating the volume average residual magnetic intensity of the material to be detected, including: performing one-dimensional Fourier transform on the obtained three-dimensional spatial signal data in three spatial directions respectively, obtaining spatial frequency data in three directions of XYZ, and synthesizing them into three-dimensional spatial frequency domain data of the material to be detected; calculating the spatial frequency coordinate range corresponding to the material surface according to the shape and size information of the material to be detected, extracting the frequency data points falling in the corresponding coordinate range from the three-dimensional spatial frequency domain data as the two-dimensional frequency domain data of the material surface; performing one-dimensional Fourier inverse transform on the two-dimensional frequency domain data in two frequency directions respectively, and reconstructing the spatial distribution nuclear magnetic resonance signal intensity corresponding to the material surface as the two-dimensional residual magnetic distribution image of the material surface; performing one-dimensional Fourier transform on the obtained one-dimensional spin echo signal data to obtain chemical shift frequency data, calculating the total nuclear magnetic resonance signal intensity corresponding to the frequency data by integration, and calculating the average nuclear magnetic resonance signal intensity per unit volume according to the volume of the material; according to the calibration relationship between the average nuclear magnetic resonance signal intensity and the residual magnetic intensity, obtaining the volume average residual magnetic intensity value of the material.

[0024] Among them, two-dimensional frequency domain data: a two-dimensional frequency domain representation obtained by Fourier transforming three-dimensional spatial signal data. In this scheme, firstly, one-dimensional Fourier transform is performed on the three-dimensional spatial signal data in the three directions of XYZ to obtain three-dimensional spatial frequency domain data. Then, according to the spatial position of the sample surface, the corresponding frequency data points are extracted from the three-dimensional spatial frequency domain data to form two-dimensional frequency domain data reflecting the nuclear spin distribution on the sample surface. The two dimensions of the two-dimensional frequency domain data correspond to the spatial frequencies in the two tangential directions of the sample surface. By performing an inverse Fourier transform on the two-dimensional frequency domain data, a two-dimensional residual magnetization distribution image of the sample surface can be reconstructed. One-dimensional inverse Fourier transform: a mathematical operation that converts a frequency domain signal into a time domain signal or a spatial domain signal. The inverse Fourier transform is the inverse process of the Fourier transform, which can synthesize the frequency components of the signal into the original signal. In this scheme, one-dimensional inverse Fourier transform is performed on the two-dimensional frequency domain data in two frequency directions, and the frequency domain information can be converted into spatial domain information to obtain a two-dimensional residual magnetization distribution image corresponding to the sample surface. The calculation formula of the one-dimensional inverse Fourier transform is that the time domain signal is equal to the frequency domain signal multiplied by the phase factor and then summed. The fast inverse Fourier transform algorithm can efficiently realize the conversion from frequency domain to spatial domain. Total NMR signal intensity: the total amplitude of the NMR signal generated by the sample. In this scheme, the one-dimensional spin echo signal data is subjected to a one-dimensional Fourier transform to obtain the chemical shift frequency data, which reflects the resonance frequency distribution of nuclear spins with different chemical shifts. By integrating the frequency data, the total NMR signal intensity contributed by all chemical shift nuclear spins can be calculated. The total NMR signal intensity is proportional to the number of nuclear spins participating in the resonance in the sample, reflecting the total remanent magnetization level of the sample. Dividing the total NMR signal intensity by the sample volume can obtain the average NMR signal intensity per unit volume, which is used for the subsequent quantitative calculation of the volume average remanent magnetization intensity of the sample. Calibration relationship between average NMR signal intensity and remanent magnetization intensity: establish a quantitative relationship between the average NMR signal intensity of the sample and its remanent magnetization intensity. Before measuring the residual magnetism intensity, it is necessary to first use a standard sample with known residual magnetism intensity to measure its average nuclear magnetic resonance signal intensity under the same experimental conditions to obtain a set of calibration data. By curve fitting the calibration data, a functional relationship between the average nuclear magnetic resonance signal intensity and the residual magnetism intensity can be established. In the subsequent measurement of unknown samples, the residual magnetism intensity value of the sample can be converted based on the measured average nuclear magnetic resonance signal intensity using the calibration relationship. The establishment of the calibration relationship requires the selection of standard samples with a material similar to that of the sample to be measured, and try to cover the residual magnetism intensity range of the sample to be measured to ensure the accuracy and reliability of the measurement results.

[0025] Specifically, S5, obtains a two-dimensional residual magnetic distribution image of the surface of the material to be tested, the image being composed of normal residual magnetic intensity data at different positions on the material surface measured by a scanning high-precision magnetometer; preprocesses the two-dimensional residual magnetic distribution image, including image denoising, image enhancement, and image segmentation, to obtain an area of ​​interest for residual magnetic distribution on the material surface; uses the Haar wavelet transform algorithm to perform a 5-level multi-scale decomposition on the image of the area of ​​interest to obtain a wavelet coefficient matrix of 1 low-frequency sub-band and 15 high-frequency sub-bands; extracts energy features from the wavelet coefficient matrix of each sub-band, calculates the 2-norm of the matrix elements, and obtains a multidimensional energy feature vector composed of 16 energy eigenvalues; selects 100 residual magnetic distribution images of the material surface with known uniformity levels, and extracts their multi-dimensional energy features. dimensional energy feature vectors as training samples for support vector machines; using RBF kernel function, grid search and cross-validation methods to optimize the hyperparameters of support vector machines, and establishing an evaluation model for the uniformity of residual magnetism distribution on the surface of materials; using the established evaluation model, the multi-dimensional energy feature vectors extracted from the residual magnetism distribution image of the surface of the material to be tested are predicted, and a uniformity score between 0 and 100 is obtained as the comprehensive uniformity index of the residual magnetism distribution on the surface of the material; according to the preset threshold interval, the comprehensive uniformity index is divided into four levels: poor, medium, good, and excellent, corresponding to the four levels of extremely weak magnetic properties of the material from poor to good; through the level judgment result of the comprehensive uniformity index, a quantitative evaluation is given on whether the material can be used in an extremely weak magnetic environment and whether the material's demagnetization process is qualified. Two artificial intelligence algorithms, wavelet transform and support vector machine, are introduced to improve the evaluation level of the uniformity of residual magnetism distribution on the surface of extremely weak magnetic materials from the perspective of signal processing and pattern recognition, making it more quantitative, intelligent and accurate.

[0026] Compared with the prior art, the advantages of this application are:

[0027] The remanent magnetism of materials is detected by using a nuclear magnetic resonance spectrometer. By utilizing the penetration and high sensitivity of nuclear magnetic resonance, comprehensive sampling of the remanent magnetism distribution on the surface and inside of the material can be achieved. By setting a three-dimensional gradient echo sequence and applying a gradient magnetic field at different spatial positions of the material, the nuclear spin precession frequency is mapped to the spatial position. The sampling signal is spatially encoded by combining fast Fourier transform to obtain a three-dimensional remanent magnetism distribution image of the material, which overcomes the defects of conventional detection methods such as many blind spots and incomplete sampling, and ensures the integrity and reliability of the evaluation of extremely weak magnetic properties.

[0028] A spin echo pulse sequence is introduced, and a 180° inversion pulse is used to phase focus nuclear spins of different chemical shifts to obtain a high signal-to-noise ratio nuclear magnetic resonance spectrum of the entire material. By optimizing the inversion pulse parameters and using a composite pulse form, the nuclear spins within the resonance peak coverage range are uniformly excited, effectively eliminating the residual phase interference caused by insufficient pulse bandwidth. At the same time, a pair of shear gradient pulses are set to suppress the influence of the internal gradient field inhomogeneity of the material, and the gradient pulse is combined to accurately encode the spin density distribution inside the material. The above technical features help to improve the quality of the sampling signal, reduce magnetic susceptibility artifacts and chemical shift artifacts, and thus ensure the accuracy of quantitative detection of extremely weak magnetic materials.

[0029] The spatial frequency domain data of the material is obtained by Fourier transforming the three-dimensional spatial signal data, and the frequency domain information of the surface is extracted according to the material size. The surface two-dimensional remanent magnetization distribution image is reconstructed by inverse Fourier transform to achieve intuitive characterization of the remanent magnetization uniformity of the material surface. The average nuclear magnetic resonance signal intensity of the material body is calculated by frequency domain analysis of the one-dimensional spin echo signal, and the average remanent magnetization intensity value of the material body is quantitatively obtained based on the calibration relationship with the remanent magnetization intensity. The above frequency domain analysis method simplifies the data processing process, and establishes a quantitative evaluation system for extremely weak magnetic properties by quantitatively calculating the remanent magnetization uniformity parameters and the average remanent magnetization intensity value, which has higher accuracy and repeatability than traditional qualitative detection methods.

[0030] The magnetization state adjustment function of the material to be tested is integrated into the nuclear magnetic resonance system. By attaching magnetization markers with known magnetization direction and strength to the surface of the material, the magnetization vector distribution of the material is determined by marker imaging, and a gradient magnetic field pulse opposite to the deviation angle is applied to align the overall magnetization direction of the material with the external static magnetic field, so as to suppress the orientation difference of the magnetization vector and enhance the sampling signal strength. Compared with conventional demagnetization and demagnetization treatment, this magnetization state adjustment method is simple to operate, can be implemented on-site, and shortens the test preparation time.

[0031] The material and tool management system is used to record the test data, and the qualification of extremely weak magnetic materials or tools is determined based on the residual magnetism uniformity and the average residual magnetism strength. The available status is automatically marked. The process-based digital management method is conducive to improving the efficiency of extremely weak magnetic material screening, reducing manual intervention, and ensuring the extremely weak magnetic performance quality of materials and tools used in construction, providing reliable technical support for high-standard construction in the fields of precision manufacturing, aerospace and military industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is an exemplary flow chart of a construction management method based on extremely weak magnetic monitoring according to some embodiments of the present application;

[0033] Figure 2 is an exemplary flow chart of residual magnetism detection according to some embodiments of the present application;

[0034] Figure 3 is an exemplary flow chart of adjusting the deviation angle of the magnetization direction according to some embodiments of the present application;

[0035] Figure 4 is an exemplary flow chart of setting a three-dimensional gradient echo pulse sequence according to some embodiments of the present application;

[0036] Figure 5 This is an exemplary flow chart of residual magnetism detection according to some embodiments of the present application. DETAILED DESCRIPTION

[0037] The method and system provided in the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0038] Figure 1 This is an exemplary flow chart of a construction management method based on extremely weak magnetic monitoring according to some embodiments of the present application. A construction management method based on extremely weak magnetic monitoring includes: obtaining extremely weak magnetic materials to be detected at the construction site. A material storage area is set up at the construction site, and the extremely weak magnetic materials entering the site are classified and stored by category, such as a non-magnetic stainless steel area and an aluminum alloy material area. The material management personnel count and register the materials in the storage area, record the name, specification, quantity, source and other information of the materials, and number the materials to generate a unique material identification code. A nuclear magnetic resonance spectrometer is used to perform a sampling residual magnetism test on the obtained extremely weak magnetic materials to be detected. According to the material category and batch, the samples to be detected are randomly selected according to a certain proportion (such as 10%) and placed in the detection cavity of the nuclear magnetic resonance spectrometer. The spectrometer uses a superconducting magnet to generate a high-uniformity static magnetic field, and is equipped with a Z-direction gradient coil and an X- and Y-direction multi-channel room temperature shim coil for applying a spatial gradient magnetic field and a radio frequency field.

[0039] Figure 2 This is an exemplary flow chart of a residual magnetization detection shown in some embodiments of the present application. During the detection, a linear gradient magnetic field parallel to the direction of the static magnetic field is first applied by the Z-direction gradient coil, so that different layers of the sample are in magnetic fields of different strengths. Then, a 90° radio frequency pulse is emitted by the X and Y-direction shim coils to excite the precession of nuclear spins at a specific layer. Next, a pair of Z-direction gradient pulses in opposite directions are applied to spatially encode the precession signal, and the encoded free induction decay signal is collected. By changing the gradient field intensity and duration in the pulse sequence, different spatial positions of the sample are sampled to obtain the residual magnetization distribution data of the sample in three-dimensional space. At the same time, a spin echo sequence is used to obtain the nuclear magnetic resonance spectrum of the entire sample. In the sequence, a 90° radio frequency pulse is first applied to excite the nuclear spin of the sample, and then a 180° inversion pulse is applied to make the nuclear spins at different positions generate echoes. The one-dimensional projection spectrum of the sample is obtained by collecting the echo signal, reflecting the average residual magnetization level of the sample.

[0040] Enter the test data into the material management system. The nuclear magnetic resonance spectrometer is connected to the material management system through a data interface to upload the collected three-dimensional spatial data and one-dimensional spectrum data. The system background extracts the spatial information of a specific area (such as the surface of the material) in the data, uses Fourier transform to obtain the frequency domain image of the residual magnetism distribution on the material surface, and calculates the residual magnetism uniformity coefficient. At the same time, the one-dimensional spectrum data is Fourier transformed to obtain the material nuclear magnetic resonance signal intensity, and the volume average residual magnetism intensity value of the material is calculated based on the calibration relationship between the signal intensity and the residual magnetism. The test results are generated in the form of a report and linked to the electronic file of the corresponding material. According to the test results, the extremely weak magnetic materials that pass the test are marked as available, otherwise they are marked as unavailable. The material management system presets material screening standards, such as the surface residual magnetism uniformity coefficient is not less than 95%, the volume average residual magnetism intensity is not higher than 1nT, etc. The single test results are compared with the screening criteria, and the materials that meet the conditions are marked as "available" in the system, and the construction personnel are allowed to use the materials. The materials that do not meet the conditions are marked as "unavailable", and their use is prohibited, and the material management personnel are notified to isolate and store them or return them.

[0041] Figure 3This is an exemplary flow chart for adjusting the deviation angle of the magnetization direction according to some embodiments of the present application. S1, place the material to be detected at the center position in the detection chamber of the nuclear magnetic resonance spectrometer, and make the magnetization direction of the material to be detected parallel to the direction of the external static magnetic field. This step includes the following sub-steps: S11, set at least three magnetization marks in different directions on the outer surface of the material to be detected. Each magnetization mark is made of a strong magnetic film material with a known magnetization direction and magnetization intensity, such as Fe, Co, Ni, etc., and the thickness is controlled within 100nm to ensure that the influence of the magnetization mark on the magnetism of the material itself is negligible. The magnetization directions of any two magnetization marks need to maintain a sufficient angle (such as greater than 30°) for subsequent calculation and distinction. The magnetization marks are pasted along different directions on the surface of the material to ensure that the amount of information on the spatial distribution of the mark is maximized. S12, use the nuclear magnetic resonance spectrometer to image each magnetization mark on the material to be detected. Adjust the radio frequency pulse frequency of the spectrometer to match the nuclear magnetic resonance frequency of the marked material to selectively excite the nuclear spins in the marked area. Frequency encoding gradient pulses and phase encoding gradient pulses are applied to spatially encode the marked area. By controlling the pulse intensity and duration, a two-dimensional projection image of the mark is obtained, and a three-dimensional mark model is reconstructed in combination with the mark thickness information. The imaging results of each mark are analyzed to obtain its spatial position (centroid coordinates), magnetization direction (long axis direction) and magnetization intensity (signal amplitude). S13, according to the spatial position, magnetization direction and magnetization intensity of each magnetization mark, a three-dimensional magnetization direction distribution model of the material to be detected is constructed. The magnetization direction inside the material volume changes continuously, with the centroid position of each mark as the node, the magnetization direction as the node direction, and the magnetization intensity as the node weight. The magnetization direction vector field distribution inside the material is fitted using methods such as three-dimensional spline interpolation. Under the premise of known material size, the magnetization direction vector field is integrated to obtain a unit vector that macroscopically characterizes the total magnetization direction of the material. S14, based on the magnetization direction vector of the material to be detected and the external static magnetic field direction vector, the angle between the two vectors is calculated. In the nuclear magnetic resonance spectrometer, the static magnetic field direction is constant and is regarded as the positive direction of the Z axis. The material magnetization direction obtained according to S13 is a unit vector of (x, y, z). Then the angle θ between the magnetization direction and the positive direction of the Z axis is arccos(z). This angle value quantifies the degree to which the material magnetization direction deviates from the external static magnetic field. S15, according to the magnetization direction deviation angle value θ, the gradient coil of the nuclear magnetic resonance spectrometer is used to apply a magnetic field gradient pulse opposite to the deviation direction, and a magnetic torque is applied to the material to be tested so that its magnetization direction gradually approaches the static magnetic field direction. Specifically, based on the material magnetization direction vector (x, y, z), let x'=-kx, y'=-ky, z'=z, where k is the proportional coefficient, which is proportional to the angle value θ. (x', y', z') is the target direction in which the material magnetization direction needs to be adjusted.Control the current input of the XYZ three-way gradient coil so that the magnetic field gradient direction generated by the coil is consistent with (x', y', z'), and the gradient strength is proportional to the magnetization strength and angle value θ of the material. Under the action of the gradient field, the overall magnetization direction of the material is rotated by the magnetic torque and moves closer to the direction of the external static magnetic field. S16, repeat steps S12 to S15, dynamically track the spatial information changes of the magnetization mark, calculate the angle value of the magnetization direction adjustment in real time, and adjust the parameters of the gradient field pulse accordingly, until the angle value θ is less than the preset angle threshold (such as 1°), it is considered that the magnetization direction of the material has been basically aligned with the direction of the external static magnetic field, and the magnetization direction adjustment process is completed. The smaller the preset angle threshold, the higher the accuracy of the magnetization direction adjustment, but it also means that the number of adjustment iterations is more, the longer the time is spent, and it needs to be reasonably set according to the actual detection requirements. S17, once the material magnetization direction adjustment meets the standard, stop applying the gradient field pulse and remove the magnetization mark on the surface of the material. The material fixing device is used to firmly fix the material to be tested that meets the magnetization direction adjustment requirements in the center of the detection chamber of the nuclear magnetic resonance spectrometer to prevent displacement or drift of the magnetization direction during subsequent detection, so as to prepare for the three-dimensional sampling of the material's residual magnetization distribution. Actively regulating the magnetization state of the material to be tested overcomes the non-uniformity of the magnetization direction caused by the magnetic anisotropy of the material itself, making the magnetization direction of the material highly consistent with the external static magnetic field. This is not only conducive to improving the signal-to-noise ratio of the three-dimensional spatial sampling of the material's residual magnetism, but also avoids the local magnetic field generated by the uneven magnetization inside the material from interfering with the nuclear magnetic resonance signal, thereby improving the sensitivity and accuracy of the material's residual magnetism detection.

[0042] Figure 4 It is an exemplary flow chart of setting a three-dimensional gradient echo pulse sequence according to some embodiments of the present application, including: S21, setting a three-dimensional gradient echo pulse sequence, the three-dimensional gradient echo pulse sequence includes a first radio frequency pulse, a first X-direction variable gradient pulse, a first Y-direction variable gradient pulse, a first Z-direction variable gradient pulse and a first sampling window. The flip angle of the first radio frequency pulse is set in the range of 30° to 90° according to the calculation result of S24. A smaller flip angle (such as 30°) can obtain a higher signal intensity, while a larger flip angle (such as 90°) can obtain more complete chemical shift information. The first radio frequency pulse adopts a Gaussian modulated soft pulse form, and the duration and bandwidth of the soft pulse are set according to the number of different chemical shifts present in the material. The duration needs to cover the frequency range of all chemical shifts, which is generally on the order of several milliseconds; the bandwidth should cover the width of the material nuclear magnetic resonance spectrum, which is generally several kilohertz. The output of the first radio frequency pulse is used to excite nuclear spins of different chemical shifts in the material to be detected to generate a transverse magnetization component as an input signal of the first sampling window.

[0043] The maximum amplitude, amplitude gradient rate and duration of the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the first Z-direction variable gradient pulse are set according to the calculation result of S23. The three variable gradient pulses adopt the form of amplitude gradient change and increasing duration. By controlling the amplitude gradient rate and the increment of duration, a magnetic field with increasing gradient is formed at different spatial positions, so that the precession frequency of the nuclear spin forms a mapping relationship with the corresponding spatial position. The maximum amplitude of the gradient pulse determines the spatial encoding resolution, the gradient rate affects the sampling bandwidth, and the duration controls the sampling coverage range, which needs to be optimized according to the sample size, the required resolution and the sampling time. The output of the three variable gradient pulses XYZ is used to perform dual encoding of the spatial position and precession frequency of the nuclear spin generated by the first radio frequency pulse excitation, forming a nuclear magnetic resonance signal with spatial position information and frequency information as the input signal of the first sampling window.

[0044] The first sampling window is set to be synchronized with the application of the first X-direction variable gradient pulse. By controlling the starting time and duration of the sampling window, a time domain NMR signal corresponding to the spatial position and precession frequency is obtained. The starting time should be after the end of the radio frequency pulse and the stability of the nuclear spin precession. The duration should cover the application process of the X-direction variable gradient pulse, and a certain margin should be reserved to prevent incomplete sampling. The first sampling window obtains the first sampling data with different spatial resolutions and signal-to-noise ratios by controlling the number of sampling points and the sampling time. The more sampling points and the longer the sampling time, the higher the spatial resolution, but the signal-to-noise ratio may decrease and the sampling time will also be extended. It is necessary to weigh the resolution and sampling efficiency and set the sampling window parameters reasonably. The collected time domain NMR signal is converted into a spatial frequency domain signal by fast Fourier transform to realize the spatial encoding of the first sampling data. Different frequency components in the frequency domain signal correspond to nuclear spin signals at different spatial positions, and the spectrum intensity reflects the content information of the material at that position.

[0045] By optimizing the various parameters of the three-dimensional gradient echo pulse sequence, including the flip angle, modulation mode, duration and bandwidth of the first RF pulse, the maximum amplitude, gradient rate and duration of the three variable gradient pulses, and the start time, duration and number of sampling points of the first sampling window, high-quality three-dimensional spatial remanent magnetization distribution data of the material can be obtained. The selection of sequence parameters must take into account factors such as material properties, required resolution, sampling efficiency, etc., and be optimized through the calculation results of S23 and S24 to meet the requirements of high-precision detection of extremely weak magnetic materials.

[0046] Specifically, taking a rare earth permanent magnet material containing three magnetic elements of Fe, Co and Ni as an example, it is known that the chemical shift range of the material is -500 to 500ppm, the spatial size is 10mm×10mm×10mm, and the required spatial resolution is 100μm×100μm×100μm. According to S23 calculation, the three-dimensional spatial sampling matrix is ​​100×100×100, the sampling field of view is 10mm×10mm×10mm, the sampling interval is 100μm, and the sampling bandwidth is 50kHz. According to S24 calculation, the flip angle of the first radio frequency pulse is set to 60°, the duration is 2ms, and the bandwidth is 5kHz. The first radio frequency pulse: adopts a soft pulse form of Gaussian modulation, with a flip angle of 60°, a duration of 2ms, and a bandwidth of 5kHz. The center frequency of the soft pulse is set to the characteristic nuclear magnetic resonance frequency of the material, such as the center frequency of Fe is 220MHz, the center frequency of Co is 213MHz, and the center frequency of Ni is 205MHz. The output power of the soft pulse is optimized according to the number of nuclear spins and relaxation time of the material, such as the output power of 500W. The output of the first RF pulse is used to simultaneously excite the nuclear spins of the three elements Fe, Co, and Ni to produce a transverse magnetization component. The first X-direction variable gradient pulse: the maximum amplitude is 50mT / m, the amplitude gradient rate is 25mT / (m·ms), the duration increases from 0.1ms to 2ms, the increment is 0.1ms, and 20 linearly increasing gradient pulses are formed. The first Y-direction variable gradient pulse: the maximum amplitude is 50mT / m, the amplitude gradient rate is 25mT / (m·ms), the duration increases from 0.1ms to 2ms, the increment is 0.1ms, and 20 linearly increasing gradient pulses are formed. The first Z-direction variable gradient pulse: the maximum amplitude is 50mT / m, the amplitude gradient rate is 25mT / (m·ms), the duration increases from 0.1ms to 2ms, the increment is 0.1ms, and 20 linearly increasing gradient pulses are formed. The three variable gradient pulses form a linearly increasing magnetic field gradient in the X, Y, and Z directions, so that the nuclear spin precession frequency at different spatial positions of the material forms a linear mapping relationship with its coordinates. The first sampling window: the number of sampling points is 1024, the sampling time is 10ms, the sampling start time is synchronized with the first X-direction variable gradient pulse, and delayed by 0.1ms to avoid the influence of pulse switching. The output of the sampling window is a time-domain nuclear magnetic resonance signal corresponding to the spatial position and precession frequency, with a signal length of 1024 points and a duration of 10ms. The sampling signal is fast Fourier transformed to obtain a spatial frequency domain signal matrix of size 100×100×100, covering a material space of 10mm×10mm×10mm, with a spatial resolution of 100μm×100μm×100μm. In the frequency domain signal matrix, each element corresponds to a voxel of 100μm×100μm×100μm in the material, and the value of the element reflects the content distribution of the three elements Fe, Co, and Ni in the voxel.By optimizing the parameters of the first RF pulse (flip angle 60°, duration 2ms, bandwidth 5kHz), the parameters of the three variable gradient pulses (maximum amplitude 50mT / m, gradient rate 25mT / (m·ms), duration 0.1-2ms), and the parameters of the first sampling window (sampling points 1024, sampling time 10ms, delay 0.1ms), a spatial resolution of 100μm×100μm×100μm can be obtained within a field of view of 10mm×10mm×10mm, and high-precision quantitative characterization of the spatial distribution of the three elements Fe, Co, and Ni inside the material can be achieved, providing accurate data support for material composition optimization and performance regulation.

[0047] S22, setting a spin echo pulse sequence, the spin echo pulse sequence includes a second radio frequency pulse, a second Z-direction variable gradient pulse, a third Z-direction variable gradient pulse, a first shear gradient pulse, a second shear gradient pulse and a second sampling window; the second radio frequency pulse is in the form of a wide-band composite pulse, including: a plurality of sub-pulses, the frequency, phase and flip angle of each sub-pulse are set according to the nuclear spin resonance frequency range of the material and the chemical shift difference, and a uniform inversion effect is obtained by changing the number and parameters of sub-pulses. The effective flip angle of the second radio frequency pulse is set within the range of 175° to 185° according to the calculation result of S24. The output of the second radio frequency pulse is used to simultaneously invert the phases of nuclear spins with different chemical shifts in the first sampling data to suppress the residual phase difference caused by insufficient inversion pulse bandwidth. The second Z-direction variable gradient pulse is applied before the second radio frequency pulse, and the output is the same as the first Z-direction variable gradient pulse. The third Z-direction variable gradient pulse is applied after the second radio frequency pulse. The application timing and amplitude change form of the second Z-direction variable gradient pulse and the third Z-direction variable gradient pulse are the same as those of the first Z-direction variable gradient pulse. The first shear gradient pulse and the second shear gradient pulse are applied before and after the second radio frequency pulse, respectively, and the amplitude and application time are calculated and set according to the distribution of magnetic field gradients between different cross sections of the material, and the gradient directions of the two shear pulses are opposite. The duration of the second sampling window is the same as that of the first sampling window.

[0048] Specifically, taking a rare earth permanent magnet material containing three magnetic elements of Fe, Co and Ni as an example, it is known that the chemical shift range of the material is -500 to 500ppm, and the chemical shifts of the three elements of Fe, Co and Ni are 0ppm, -200ppm and 300ppm respectively. The material size is 10mm×10mm×10mm, and the magnetic field gradient between different cross sections is 0.1mT / mm. According to S24 calculation, the effective flip angle of the second radio frequency pulse is set to 180°. The second radio frequency pulse: adopts the form of a wideband composite pulse, including 5 sub-pulses, each sub-pulse has a duration of 0.4ms, and the total duration is 2ms. The frequencies of the 5 sub-pulses are set to 220MHz, 213MHz, 205MHz, 227MHz and 198MHz, respectively, corresponding to the chemical shifts of Fe, Co, Ni and the frequencies at -300ppm and 500ppm, covering the entire chemical shift range of the material. The phases of the five sub-pulses were set to 0°, 90°, 180°, 270° and 0°, respectively. The synchronization between sub-pulses of different frequencies was enhanced by changing the phases. The flip angles of the five sub-pulses were set to 36°, 36°, 36°, 36° and 36°, respectively, totaling 180°, which met the requirements of the effective flip angle. The output power of the composite pulse was 500W, which was used to simultaneously reverse the phases of the nuclear spins of the three elements Fe, Co and Ni. The second Z-direction variable gradient pulse: the maximum amplitude was 50mT / m, the amplitude gradient rate was 25mT / (m·ms), the duration increased from 0.1ms to 2ms, and the increment was 0.1ms, forming 20 linearly increasing gradient pulses, which were the same as the first Z-direction variable gradient pulse and were applied before the second RF pulse. The third Z-direction variable gradient pulse: the maximum amplitude is 50mT / m, the amplitude gradient rate is 25mT / (m·ms), the duration increases from 0.1ms to 2ms, the increment is 0.1ms, forming 20 linearly increasing gradient pulses, which are the same as the first Z-direction variable gradient pulse and are applied after the second RF pulse. The two Z-direction variable gradient pulses are used to form the same spatial encoding in the Z direction as the first sampling. The first shear gradient pulse: applied before the second RF pulse, the maximum amplitude is 10mT / m, and the duration is 2ms. The second shear gradient pulse: applied after the second RF pulse, the maximum amplitude is -10mT / m, and the duration is 2ms. The amplitude and duration of the two shear gradient pulses are calculated based on the magnetic field gradient of 0.1mT / mm between different sections of the material, which is used to compensate for the magnetic field inhomogeneity between different sections, so that the inversion pulse has the same flipping effect on all sections. The second sampling window: the number of sampling points is 1024, and the sampling time is 10ms, which is the same as the first sampling window. The output of the second sampling window is the time domain nuclear magnetic resonance signal of the three elements Fe, Co and Ni after the nuclear spins are reversed. The signal length is 1024 points and the duration is 10ms.The sampling signal is subjected to fast Fourier transform to obtain a spatial frequency domain signal matrix of size 100×100×100, covering a material space of 10mm×10mm×10mm, with a spatial resolution of 100μm×100μm×100μm. The frequency domain signal matrix reflects the spatial distribution of the nuclear spin reversal of the three elements Fe, Co, and Ni inside the material. By optimizing the parameters of the second RF pulse (the composite pulse contains 5 sub-pulses, frequencies of 220 / 213 / 205 / 227 / 198 MHz, phases of 0° / 90° / 180° / 270° / 0°, flip angles of 36°×5, and durations of 0.4 ms×5), the parameters of the two Z-direction variable gradient pulses (maximum amplitude of 50 mT / m, gradient rate of 25 mT / (m·ms), and durations of 0.1-2 ms), the parameters of the two shear gradient pulses (the first shear pulse is 10 mT / m, 2 ms, and the second shear pulse is -10 mT / m, 2 ms), and the parameters of the second sampling window (sampling points of 1024, sampling time of 10 ms), the spatial distribution data of Fe, Co, and Ni obtained by the first sampling can be inverted and phase corrected, artifacts and distortions caused by chemical shift and magnetic field inhomogeneity can be eliminated, the quantitative accuracy of material nuclear magnetic resonance imaging can be improved, and reliable image data can be provided for subsequent material evaluation.

[0049] S23, according to the magnetic susceptibility anisotropy of the material to be detected and the required spatial resolution of the residual magnetization distribution, the maximum amplitude, amplitude gradient rate and duration of the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the first Z-direction variable gradient pulse are calculated; S24, according to the relaxation time constant T1 and the relaxation time constant T2 of the material to be detected, the flip angles of the first radio frequency pulse and the second radio frequency pulse are calculated, and the calculation results are used to set the flip angles of the two radio frequency pulses in S21 and S22 respectively, so as to reduce the signal attenuation caused by longitudinal and transverse relaxation while satisfying the nuclear spin excitation and inversion; at the same time, the time interval between the second radio frequency pulse and the first radio frequency pulse is calculated, and the calculation result is used to set the application time of the second radio frequency pulse in S22, so as to avoid the first sampling data signal from decaying to the noise level before the phase inversion due to insufficient relaxation time constant T2.

[0050] Specifically, taking a rare earth permanent magnet material containing three magnetic elements of Fe, Co and Ni as an example, the magnetic susceptibility anisotropy parameters of the material are known to be: X-direction magnetic susceptibility χx=0.05, Y-direction magnetic susceptibility χy=0.08, and Z-direction magnetic susceptibility χz=0.12. The relaxation time constant T1 of the material is 500ms, and the relaxation time constant T2 is 50ms. The required spatial resolution of the residual magnetization distribution is 100μm×100μm×100μm.

[0051] S23: According to the material magnetic susceptibility anisotropy parameters χx, χy, χz, calculate the magnetic field gradient requirements in the three directions of X, Y, and Z. Since χz>χy>χx, the magnetic field gradient in the Z direction is the largest, followed by the Y direction, and the smallest in the X direction. The magnetic field gradient required in the Z direction is Gz, then the magnetic field gradient in the Y direction Gy=Gz×(χy / χz)=Gz×0.67, and the magnetic field gradient in the X direction Gx=Gz×(χx / χz)=Gz×0.42. According to the required spatial resolution of 100μm and the material size of 10mm, the three-dimensional spatial sampling matrix is ​​calculated to be 100×100×100, the sampling field of view FOV is 10mm×10mm×10mm, and the sampling bandwidth SW is 50kHz. From the sampling theory, it can be seen that the relationship between the magnetic field gradient G and the sampling bandwidth SW and the sampling field of view FOV is: G=2π×SW / (γ×FOV), where γ is the gyromagnetic ratio, which is 42.58MHz / T for protons. Substituting the data, we can get the magnetic field gradient Gz in the Z direction = 59.4mT / m, the magnetic field gradient Gy in the Y direction = 39.6mT / m, and the magnetic field gradient Gx in the X direction = 24.8mT / m. To meet the spatial resolution requirements, the maximum amplitudes of the three variable gradient pulses are set as follows: 24.8mT / m for the first X-direction variable gradient pulse, 39.6mT / m for the first Y-direction variable gradient pulse, and 59.4mT / m for the first Z-direction variable gradient pulse. To cover the entire sampling field of view, the duration of the variable gradient pulse should not be less than the sampling time, which is set to 10ms. To make the gradient field change linearly in the sample, the amplitude change rate of the variable gradient pulse is set as follows: 2.48mT / (m·ms) for the first X-direction variable gradient pulse, 3.96mT / (m·ms) for the first Y-direction variable gradient pulse, and 5.94mT / (m·ms) for the first Z-direction variable gradient pulse.

[0052] Figure 5This is an exemplary flow chart of residual magnetization detection according to some embodiments of the present application. According to the material relaxation time constants T1 and T2, the flip angles of the first radio frequency pulse and the second radio frequency pulse are calculated. For a gradient echo sequence, in order to obtain the maximum signal strength, the flip angle of the first radio frequency pulse is generally set to the Ernst angle, i.e., arccos(exp(-TR / T1)), where TR is the repetition time, which is generally less than T1. TR=200ms, then the flip angle of the first radio frequency pulse is 76°. For a spin echo sequence, the flip angle of the second radio frequency pulse is generally set to 180° to achieve complete spin reversal. Therefore, the flip angle of the first radio frequency pulse is set to 76° in S21, and the flip angle of the second radio frequency pulse is set to 180° in S22. According to the material relaxation time constant T2, the time interval TE between the second radio frequency pulse and the first radio frequency pulse is calculated. Since the signal intensity of the first sampling data is proportional to exp(-TE / T2), in order to avoid the signal from attenuating to the noise level, TE≤T2 / 10 is generally taken. T2=50ms, then TE should not be greater than 5ms. Taking into account the duration of the radio frequency pulse and the variable gradient pulse, the application time of the second radio frequency pulse is set in S22 to 3ms after the end of the first radio frequency pulse. At this time, the signal intensity of the first sampling data is about 94% of the initial value, which can meet the needs of spin reversal and spatial encoding. Under the above parameter settings, a high-resolution image of the three-dimensional spatial remanent magnetization distribution of rare earth permanent magnet materials can be obtained. The first radio frequency pulse flip angle is 76°, the second radio frequency pulse flip angle is 180°, and the second radio frequency pulse application time is 3ms after the end of the first radio frequency pulse, which can meet the nuclear spin excitation and reversal while reducing the signal attenuation caused by longitudinal and transverse relaxation. The maximum amplitude of the first X-direction variable gradient pulse is 24.8mT / m, the gradient rate is 2.48mT / (m·ms), and the duration is 10ms. The maximum amplitude of the first Y-direction variable gradient pulse is 39.6mT / m, the gradient rate is 3.96mT / (m·ms), and the duration is 10ms. The maximum amplitude of the first Z-direction variable gradient pulse is 59.4mT / m, the gradient rate is 5.94mT / (m·ms), and the duration is 10ms. It can achieve a spatial resolution of 100μm×100μm×100μm within a field of view of 10mm×10mm×10mm, meeting the spatial encoding requirements of material magnetic susceptibility anisotropy.

[0053] S3, first execute the set three-dimensional gradient echo pulse sequence, collect nuclear magnetic resonance signals, and obtain three-dimensional spatial signal data of the material to be detected. According to the parameters set in S21, a three-dimensional gradient echo pulse sequence is executed, including applying the first radio frequency pulse, the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse, the first Z-direction variable gradient pulse and the first sampling window. In the first sampling window, the nuclear magnetic resonance signal is collected to obtain the first sampling data as the three-dimensional spatial signal data of the material to be detected. The first sampling data is a three-dimensional complex matrix of size Nx×Ny×Nz, where Nx, Ny, and Nz are the number of sampling points in the three directions of X, Y, and Z, respectively. The first sampling data reflects the distribution of nuclear spins at different spatial positions on the surface and inside of the material to be detected. By performing Fourier transform on the first sampling data, a three-dimensional spatial distribution image of the magnetic elements inside the material can be obtained, and the spatial resolution reaches the result calculated by S23.

[0054] After the three-dimensional gradient echo pulse sequence is executed, the set spin echo pulse sequence is executed to collect nuclear magnetic resonance signals and obtain one-dimensional spin echo signal data of the material to be detected. According to the parameters set in S22, a spin echo pulse sequence is executed, including applying a second radio frequency pulse, a second Z-direction variable gradient pulse, a third Z-direction variable gradient pulse, a first shear gradient pulse, a second shear gradient pulse and a second sampling window. In the second sampling window, the nuclear magnetic resonance signal is collected to obtain the second sampling data as the one-dimensional spin echo signal data of the material to be detected. The second sampling data is a one-dimensional complex vector with a length of Nz. The second sampling data reflects the resonant frequency distribution of the overall nuclear spin of the material to be detected after the spin echo. By Fourier transforming the second sampling data, the spectral distribution of different chemical shift nuclear spins inside the material can be obtained, and the frequency resolution reaches the range of the material nuclear spin resonance frequency. The first sampling data and the second sampling data are used in combination to achieve the simultaneous characterization of the spatial distribution and chemical shift distribution of magnetic elements inside the material. The first sampling data provides the three-dimensional spatial distribution information of the magnetic elements inside the material, but due to the limitations of the sampling bandwidth and the RF pulse bandwidth, the signals of nuclear spins with different chemical shifts may overlap, affecting the spatial resolution. The second sampling data provides the spectral distribution information of the nuclear spins of the entire material. By identifying and quantifying the signals of nuclear spins with different chemical shifts, the relative content of each magnetic element can be calculated. The spatial information in the first sampling data is combined with the spectral information in the second sampling data to locate and quantify the signals of nuclear spins with different chemical shifts, thereby obtaining a high-resolution three-dimensional image of the spatial distribution and chemical shift distribution of the magnetic elements inside the material.

[0055] Specifically, taking a rare earth permanent magnet material containing three magnetic elements of Fe, Co and Ni as an example, the known material size is 10mm×10mm×10mm. According to the calculation results of S23 and S24, the specific parameters of the three-dimensional gradient echo pulse sequence and the spin echo pulse sequence are set as follows: three-dimensional gradient echo pulse sequence: first radio frequency pulse: flip angle 76°, duration 0.5ms, first X-direction variable gradient pulse: maximum amplitude 24.8mT / m, gradient rate 2.48mT / (m·ms), duration 10ms, first Y-direction variable gradient pulse: maximum amplitude 39.6mT / m, gradient rate 3.96mT / (m·ms), duration 10ms. First Z-direction variable gradient pulse: maximum amplitude 59.4mT / m, gradient rate 5.94mT / (m·ms), duration 10ms, first sampling window: duration 5ms, number of sampling points Nx=Ny=Nz=100.

[0056] Spin echo pulse sequence: second radio frequency pulse: the composite pulse consists of 5 sub-pulses, with a total flip angle of 180° and a duration of 1ms. The second Z-direction variable gradient pulse: maximum amplitude 59.4mT / m, gradient rate 5.94mT / (m·ms), duration 10ms, applied before the second radio frequency pulse. The third Z-direction variable gradient pulse: maximum amplitude 59.4mT / m, gradient rate 5.94mT / (m·ms), duration 10ms, applied after the second radio frequency pulse. The first shear gradient pulse: maximum amplitude 20mT / m, duration 2ms, applied before the second radio frequency pulse. The second shear gradient pulse: maximum amplitude -20mT / m, duration 2ms, applied after the second radio frequency pulse. The second sampling window: duration 5ms, number of sampling points Nf = 1024.

[0057] According to the above parameters, the operation step of S3 is performed: a three-dimensional gradient echo pulse sequence is executed once, the nuclear magnetic resonance signal in the first sampling window is collected, the first sampling data D1 is obtained, and a three-dimensional complex matrix of 100×100×100 is formed. The spatial distribution of the three elements Fe, Co, and Ni in the material is shown in the figure below, where Fe is mainly distributed in the central area, Co is mainly distributed in the edge area, and Ni is evenly distributed throughout the material. The first sampling data D1 reflects this spatial distribution feature, but due to the limitation of the sampling bandwidth (50kHz), the signals of the three elements overlap to a certain extent. A spin echo pulse sequence is executed once, the nuclear magnetic resonance signal in the second sampling window is collected, and the second sampling data D2 is obtained, forming a one-dimensional complex vector of 1×1024. The resonance frequencies of the three elements Fe, Co, and Ni are 200MHz, 205MHz, and 210MHz, respectively. The second sampling data D2 can distinguish the signals of the three elements, and the spectrum distribution shown in the figure below is obtained, where the signal peak area ratio of Fe, Co, and Ni is approximately 5:3:2, reflecting the relative content of the three elements. The first sampling data D1 and the second sampling data D2 are registered and fused to obtain a high-resolution three-dimensional image of the spatial distribution and relative content of the three elements inside the material, as shown in the figure below. Different colors in the image represent different elements, and the depth of the color represents the relative content of the elements. The spatial resolution reaches 100μm, and the quantitative results of the three elements are consistent with the spectrum analysis. Through the above-mentioned S3 embodiment, the spatial signal data D1 and the spectrum signal data D2 of the material are respectively obtained by using two pulse sequences of three-dimensional gradient echo and spin echo, and the complete information of the distribution and content of magnetic elements inside the material is obtained by data fusion. Compared with the method of using three-dimensional gradient echo or spin echo alone, this method can obtain higher spatial resolution and spectrum resolution, and realize quantitative analysis of element content, which can be used for component optimization and structural characterization of rare earth permanent magnet materials.

[0058] S4, performing one-dimensional Fourier transform on the obtained three-dimensional spatial signal data in three spatial directions respectively, obtaining spatial frequency data in three directions of XYZ, and synthesizing them into three-dimensional spatial frequency domain data of the material to be detected; calculating the spatial frequency coordinate range corresponding to the material surface according to the shape and size information of the material to be detected, extracting the frequency data points falling in the corresponding coordinate range from the three-dimensional spatial frequency domain data as the two-dimensional frequency domain data of the material surface; performing one-dimensional inverse Fourier transform on the two-dimensional frequency domain data in two frequency directions respectively, reconstructing the spatial distribution nuclear magnetic resonance signal intensity corresponding to the material surface as the two-dimensional residual magnetization distribution image of the material surface; performing one-dimensional Fourier transform on the obtained one-dimensional spin echo signal data to obtain chemical shift frequency data, calculating the total nuclear magnetic resonance signal intensity corresponding to the frequency data by integration, and calculating the average nuclear magnetic resonance signal intensity per unit volume according to the volume of the material; obtaining the volume average residual magnetization intensity value of the material according to the calibration relationship between the average nuclear magnetic resonance signal intensity and the residual magnetization intensity.

[0059] Specifically, taking a rare earth permanent magnetic material with a size of 10mm×10mm×1mm as an example, combined with the first sampling data D1 and the second sampling data D2 obtained in the S3 embodiment, the embodiment of S4 is specifically explained as follows: the first sampling data D1 (with a size of 100×100×100) is subjected to a one-dimensional Fourier transform in the three spatial directions of X, Y, and Z, respectively, to obtain spatial frequency data Dx, Dy, and Dz in three directions, each of which has a size of 100×100×100. Dx, Dy, and Dz are multiplied according to the frequency values ​​of the corresponding positions to obtain three-dimensional spatial frequency domain data D3, which reflects the spatial frequency distribution characteristics of the magnetic elements inside the material. According to the shape (rectangle) and size (10mm×10mm×1mm) of the material, the spatial frequency coordinate range corresponding to the upper surface of the material is calculated as follows: X-direction frequency range: [0, 10]m^(-1), Y-direction frequency range: [0, 10]m^(-1), Z-direction frequency range: [99, 100]m^(-1), and the frequency data points falling in the above coordinate range are extracted from the three-dimensional spatial frequency domain data D3 to form a 100×100 two-dimensional frequency domain data D4, which reflects the frequency distribution characteristics of the upper surface of the material. The two-dimensional frequency domain data D4 is subjected to a one-dimensional Fourier inverse transform in the X and Y frequency directions to obtain a 100×100 two-dimensional spatial domain data D5, which represents the nuclear magnetic resonance signal intensity distribution on the upper surface of the material. D5 is scaled according to the actual size of the material (10mm×10mm) to obtain a two-dimensional residual magnetization distribution image of the upper surface of the material, as shown in the figure below. Different colors in the image represent the level of residual magnetism. It can be seen that the residual magnetism on the surface of the material is unevenly distributed, and the residual magnetism in the central area is significantly higher than that in the edge area. The second sampling data D2 (size is 1×1024) is subjected to a one-dimensional Fourier transform to obtain chemical shift frequency data Df, size is 1×1024. According to the spectrum analysis results in the S3 embodiment, the three frequency points of 200MHz, 205MHz and 210MHz in Df correspond to the signal peaks of the three elements Fe, Co and Ni respectively. The data points of the three signal peaks in Df are integrated respectively, and the total intensity of the nuclear magnetic resonance signals of the three elements Fe, Co and Ni are obtained as S_Fe=5000, S_Co=3000 and S_Ni=2000 respectively.

[0060] According to the volume V = 100mm^3 of the material, the average nuclear magnetic resonance signal intensity per unit volume of the three elements Fe, Co, and Ni is calculated: I_Fe = S_Fe / V = 50a.u. / mm^3, I_Co = S_Co / V = 30a.u. / mm^3, I_Ni = S_Ni / V = 20a.u. / mm^3, where au represents an arbitrary unit. It is known that the calibration relationship between the nuclear magnetic resonance signal intensity and the residual magnetic intensity of the three elements Fe, Co, and Ni is as follows: 1a.u. / mm^3->10mT, then the volume average residual magnetic intensity of the three elements Fe, Co, and Ni of the material can be calculated as: B_Fe = 500mT, B_Co = 300mT, B_Ni = 200mT, and the overall average residual magnetic intensity of the material is B = B_Fe + B_Co + B_Ni = 1000mT.

[0061] Taking the two-dimensional residual magnetization distribution image (100×100 pixels, resolution 0.1 mm) of the upper surface of the rare earth permanent magnet material obtained in the S4 embodiment as an example, the embodiment of S5 is specifically explained as follows: the two-dimensional residual magnetization distribution image is preprocessed, including: using a 5×5 Gaussian filter to denoise the image, smoothing the mutation of the residual magnetization intensity, using a histogram equalization algorithm to enhance the image, improving the contrast of the residual magnetization distribution, using the Otsu threshold segmentation algorithm to binarize the image, and extracting the region of interest (ROI) with a size of 80×80 pixels. The ROI image is decomposed at 5 levels of multi-scale using the Haar wavelet transform algorithm to obtain: 1 80×80 low-frequency sub-band LL5, 5 80×80 horizontal high-frequency sub-bands LH1~LH5, 5 80×80 vertical high-frequency sub-bands HL1~HL5, 5 80×80 diagonal high-frequency sub-bands HH1~HH5, a total of 16 wavelet coefficient matrices. Energy features are extracted from the 16 wavelet coefficient matrices respectively, and the 2-norm of the matrix elements is calculated: E_i = sqrt(sum(C_i.^2)), i = 1, 2, ..., 16, where C_i represents the i-th wavelet coefficient matrix and E_i represents its corresponding energy eigenvalue. The 16 energy eigenvalues ​​are combined into a multidimensional energy eigenvector E = [E_1, E_2, ..., E_16]. The multidimensional energy feature vector extracted from the residual magnetization distribution image of the material surface is: E = [120.3, 23.5, 31.2, 12.8, 15.6, 27.1, 20.9, 14.3, 11.7, 21.4, 18.2, 9.6, 8.4, 16.7, 13.5, 7.2]. 100 known uniformity levels (13) are selected to extract their multidimensional energy feature vectors to form a 100×16 training sample matrix X_train. The corresponding uniformity levels are formed into a 100×1 target value vector Y_train. The RBF kernel function is used to build a support vector machine regression model, and grid search and 5 The model hyperparameters were optimized by the cross-validation method: penalty coefficient C: search range is [0.1, 1, 10, 100], the optimal value is 10, RBF kernel function parameter gamma: search range is [0.01, 0.1, 1, 10], the optimal value is 0.1. The support vector machine model was trained using the optimal hyperparameter combination (C = 10, gamma = 0.1) to obtain an evaluation model for the uniformity of the residual magnetic distribution on the material surface. The evaluation model was used to predict the multidimensional energy feature vector E extracted from the residual magnetic distribution image on the material surface, and the uniformity score was S = 72.5, indicating that the comprehensive uniformity index of the residual magnetic distribution on the material surface was 72.5.

[0062] The comprehensive uniformity index is divided into four levels according to the preset threshold range: 0<=S<25: poor, the material's extremely weak magnetic properties are very poor and cannot be used in extremely weak magnetic environments; 25<=S<50: medium, the material's extremely weak magnetic properties are poor and the demagnetization process is unqualified; 50<=S<75: good, the material's extremely weak magnetic properties are good and the demagnetization process is basically qualified; 75<=S<=100: excellent, the material's extremely weak magnetic properties are excellent and the demagnetization process is completely qualified. In this case, the comprehensive uniformity index S=72.5, which belongs to the "good" level, indicating that the rare earth permanent magnet material has good extremely weak magnetic properties, its surface residual magnetism distribution is relatively uniform, the demagnetization process is basically qualified, and it can be used in general extremely weak magnetic environments, but there is still room for further improvement.

Claims

1. A construction management method based on extremely weak magnetic monitoring, comprising: Obtain extremely weak magnetic materials to be tested at the construction site; wherein the extremely weak magnetic materials include non-magnetic stainless steel and aluminum alloy; A nuclear magnetic resonance spectrometer is used to perform a sampling residual magnetism test on the extremely weak magnetic material to be tested; and the test data is entered into the material management system; wherein the nuclear magnetic resonance spectrometer includes a Z-direction gradient coil, and a multi-channel room temperature shim coil in the X-direction and the Y-direction; According to the test results, the extremely weak magnetic materials that pass the test are marked as usable for construction use; otherwise, they are marked as unusable; The nuclear magnetic resonance spectrometer is used to perform sampling residual magnetism detection on the extremely weak magnetic materials to be detected, including: S1, placing the material to be tested at the center of the detection chamber of the nuclear magnetic resonance spectrometer, and making the magnetization direction of the material to be tested parallel to the direction of the external static magnetic field; S2, setting a three-dimensional gradient echo pulse sequence and a spin echo pulse sequence on the nuclear magnetic resonance spectrometer; S3, successively executing the set three-dimensional gradient echo pulse sequence and spin echo pulse sequence to respectively collect nuclear magnetic resonance signals and obtain three-dimensional spatial signal data and one-dimensional spin echo signal data of the material to be detected; S4, performing a three-dimensional Fourier transform on the obtained three-dimensional spatial signal data of the material to be detected to obtain three-dimensional spatial frequency domain data of the material to be detected; extracting two-dimensional frequency domain data corresponding to the surface of the material to be detected from the three-dimensional spatial frequency domain data according to the geometric size of the material to be detected; performing an inverse Fourier transform on the extracted two-dimensional frequency domain data to reconstruct a two-dimensional residual magnetic distribution image of the surface of the material to be detected; performing a one-dimensional Fourier transform on the obtained one-dimensional spin echo signal data to obtain one-dimensional frequency domain data of the material to be detected, and calculating the volume average residual magnetic intensity of the material to be detected according to the one-dimensional frequency domain data; S5, respectively calculating the uniformity parameters of the reconstructed two-dimensional residual magnetization distribution image of the surface of the material to be detected and the volume average residual magnetization intensity of the material to be detected; judging whether the material to be detected meets the requirements of the extremely weak magnetic material according to the uniformity of the surface residual magnetization distribution and the volume average residual magnetization intensity; if the uniformity of the surface residual magnetization distribution is higher than the first preset threshold value and the volume average residual magnetization intensity is lower than the second preset threshold value, it is judged as a qualified extremely weak magnetic material, otherwise it is unqualified.

2. The construction management method based on extremely weak magnetic monitoring according to claim 1 is characterized in that: S1, placing the material to be tested directly at the center of the detection chamber of the nuclear magnetic resonance spectrometer, and making the magnetization direction of the material to be tested parallel to the direction of the external static magnetic field, including: S11, setting at least three magnetization marks with different directions on the outer surface of the material to be detected, each magnetization mark is made of a strong magnetic film material with a known magnetization direction and magnetization intensity, and the magnetization directions of any two magnetization marks are not parallel; S12, using a nuclear magnetic resonance spectrometer to image each magnetic marker on the material to be detected, and obtaining the spatial position, magnetization direction and magnetization intensity of each magnetic marker; S13, constructing a three-dimensional magnetization direction distribution model of the material to be detected according to the spatial position, magnetization direction and magnetization intensity of each magnetization mark, and calculating the magnetization direction vector of the material to be detected; S14, based on the magnetization direction vector of the material to be tested and the direction vector of the external static magnetic field, calculating the angle between the two vectors to obtain the angle value of the magnetization direction deviating from the static magnetic field direction; S15, according to the deviation angle value of the magnetization direction, using the gradient coil of the nuclear magnetic resonance spectrometer to apply a magnetic field gradient pulse opposite to the deviation direction, and applying a magnetic torque to the material to be tested, so that its magnetization direction gradually approaches the static magnetic field direction; S16, repeating steps S12 to S15 until the calculated magnetization direction deviation angle value is less than a preset angle threshold; S17, stop applying the magnetic field gradient pulse, and fix the material to be detected that meets the magnetization direction adjustment requirement at the center position in the detection cavity of the nuclear magnetic resonance spectrometer.

3. The construction management method based on extremely weak magnetic monitoring according to claim 1 is characterized in that: S2, setting a three-dimensional gradient echo pulse sequence and a spin echo pulse sequence on a nuclear magnetic resonance spectrometer, including: S21, setting a three-dimensional gradient echo pulse sequence, wherein the three-dimensional gradient echo pulse sequence comprises a first radio frequency pulse, a first X-direction variable gradient pulse, a first Y-direction variable gradient pulse, a first Z-direction variable gradient pulse and a first sampling window; S22, setting a spin echo pulse sequence, wherein the spin echo pulse sequence includes a second radio frequency pulse, a second Z-direction variable gradient pulse, a third Z-direction variable gradient pulse, a first shear gradient pulse, a second shear gradient pulse, and a second sampling window; S23, calculating the maximum amplitude, amplitude gradient rate and duration of the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the first Z-direction variable gradient pulse according to the magnetic susceptibility anisotropy of the material to be detected and the required spatial resolution of the residual magnetization distribution; S24, calculating the flip angles of the first RF pulse and the second RF pulse according to the relaxation time constant T1 and the relaxation time constant T2 of the material to be detected, and the calculation results are used to set the flip angles of the two RF pulses in S21 and S22, respectively, so as to reduce the signal attenuation caused by longitudinal and transverse relaxation while satisfying the nuclear spin excitation and inversion; at the same time, calculating the time interval between the second RF pulse and the first RF pulse, and the calculation result is used to set the application time of the second RF pulse in S22, so as to avoid the first sampling data signal from decaying to the noise level before the phase inversion due to insufficient relaxation time constant T2.

4. The construction management method based on extremely weak magnetic monitoring according to claim 3 is characterized in that: S21, setting a three-dimensional gradient echo pulse sequence, the three-dimensional gradient echo pulse sequence comprising a first radio frequency pulse, a first X-direction variable gradient pulse, a first Y-direction variable gradient pulse, a first Z-direction variable gradient pulse and a first sampling window, including: The flip angle of the first radio frequency pulse is set in the range of 30° to 90° according to the calculation result of S24; the output of the first radio frequency pulse is used to excite nuclear spins of different chemical shifts in the material to be detected to generate a transverse magnetization component as an input signal of the first sampling window; The maximum amplitude, amplitude gradient rate and duration of the first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the first Z-direction variable gradient pulse are set according to the calculation result of S23, and the outputs of the three XYZ variable gradient pulses are used to perform dual encoding of the spatial position and precession frequency of the nuclear spins generated by the first radio frequency pulse excitation, so as to form a nuclear magnetic resonance signal with spatial position information and frequency information as the input signal of the first sampling window; The first sampling window obtains first sampling data with different spatial resolutions and signal-to-noise ratios by controlling the number of sampling points and sampling time.

5. The construction management method based on extremely weak magnetic monitoring according to claim 4 is characterized in that: S21, the output of the three variable gradient pulses XYZ is used to perform dual encoding of the spatial position and precession frequency of the nuclear spins generated by the first radio frequency pulse excitation, including: The first radio frequency pulse adopts the form of a Gaussian modulated soft pulse, and the duration and bandwidth of the soft pulse are set according to the number of different chemical shifts existing in the material; The first X-direction variable gradient pulse, the first Y-direction variable gradient pulse and the first Z-direction variable gradient pulse adopt the form of gradually changing amplitude and increasing duration, and by controlling the rate of gradually changing amplitude and the increment of duration, a magnetic field with increasing gradient is formed at different spatial positions, so that the precession frequency of the nuclear spin forms a mapping relationship with the corresponding spatial position; The first sampling window is set to be synchronized with the application of the first X-direction variable gradient pulse. By controlling the starting time and duration of the sampling window, a time domain nuclear magnetic resonance signal corresponding to the spatial position and the precession frequency is obtained, and is converted into a spatial frequency domain signal through a fast Fourier transform to spatially encode the first sampling data.

6. The construction management method based on extremely weak magnetic monitoring according to claim 3 is characterized in that: S22, setting a spin echo pulse sequence, the spin echo pulse sequence comprising a second radio frequency pulse, a second Z-direction variable gradient pulse, a third Z-direction variable gradient pulse, a first shear gradient pulse, a second shear gradient pulse and a second sampling window, including: The effective flip angle of the second radio frequency pulse is set in the range of 175° to 185° according to the calculation result of S24, and the output of the second radio frequency pulse is used to simultaneously invert the phases of nuclear spins with different chemical shifts in the first sampling data to suppress the residual phase difference caused by insufficient inversion pulse bandwidth; The second Z-direction variable gradient pulse is applied before the second radio frequency pulse, and the output is the same as the first Z-direction variable gradient pulse; The third Z-direction variable gradient pulse is applied after the second radio frequency pulse; the application timing and amplitude variation form of the second Z-direction variable gradient pulse and the third Z-direction variable gradient pulse are the same as those of the first Z-direction variable gradient pulse; The first shear gradient pulse and the second shear gradient pulse are applied before and after the second radio frequency pulse respectively, and the amplitude and application time are calculated and set according to the distribution of magnetic field gradients between different cross sections of the material, and the gradient directions of the two shear pulses are opposite; The duration of the second sampling window is the same as that of the first sampling window.

7. The construction management method based on extremely weak magnetic monitoring according to claim 6 is characterized in that: The second radio frequency pulse is in the form of a wideband composite pulse, including: multiple sub-pulses, the frequency, phase and flip angle of each sub-pulse are set according to the nuclear spin resonance frequency range and chemical shift difference of the material, and a uniform inversion effect is obtained by changing the number and parameters of sub-pulses.

8. The construction management method based on extremely weak magnetic monitoring according to claim 7 is characterized in that: S3, successively executing the set three-dimensional gradient echo pulse sequence and spin echo pulse sequence, respectively collecting nuclear magnetic resonance signals, and obtaining three-dimensional spatial signal data and one-dimensional spin echo signal data of the material to be tested, including: According to the set three-dimensional gradient echo pulse sequence, a three-dimensional gradient echo pulse sequence is executed once, and first sampling data obtained in a first sampling window is collected as three-dimensional spatial signal data of the material to be detected; According to the set spin echo pulse sequence, after the three-dimensional gradient echo pulse sequence is executed, a spin echo pulse sequence is executed once to collect second sampling data obtained in the second sampling window as one-dimensional spin echo signal data of the material to be detected; The first sampling data reflects the distribution of nuclear spins at different spatial positions on the surface and inside of the material to be detected; the second sampling data reflects the resonant frequency distribution of the overall nuclear spin of the material to be detected after spin echo.

9. The construction management method based on extremely weak magnetic monitoring according to claim 8 is characterized in that: S4, calculating the volume average residual magnetic intensity of the material to be tested, including: Perform one-dimensional Fourier transform on the acquired three-dimensional spatial signal data in three spatial directions respectively to obtain spatial frequency data in three directions of XYZ, and synthesize them into three-dimensional spatial frequency domain data of the material to be tested; According to the shape and size information of the material to be tested, the spatial frequency coordinate range corresponding to the material surface is calculated, and the frequency data points falling in the corresponding coordinate range are extracted from the three-dimensional spatial frequency domain data as the two-dimensional frequency domain data of the material surface; Perform one-dimensional inverse Fourier transform on the two-dimensional frequency domain data in two frequency directions respectively, and reconstruct the spatial distribution nuclear magnetic resonance signal intensity corresponding to the material surface as a two-dimensional residual magnetic distribution image of the material surface; The obtained one-dimensional spin echo signal data is subjected to one-dimensional Fourier transformation to obtain chemical shift frequency data, the total nuclear magnetic resonance signal intensity corresponding to the frequency data is calculated by integration, and the average nuclear magnetic resonance signal intensity per unit volume is calculated according to the volume of the material; According to the calibration relationship between the average nuclear magnetic resonance signal intensity and the residual magnetic intensity, the volume average residual magnetic intensity value of the material is obtained.

Images