Ferromagnetic workpiece mechanical vibration measurement system and method based on acfm

Abstract

This invention discloses an ACFM-based mechanical vibration measurement system and method for ferromagnetic workpieces, belonging to the field of mechanical vibration measurement technology. The method includes: applying a rotating alternating electromagnetic field excitation; adjusting the magnetic field uniformity in a closed loop; having a sensor array vibrate with the workpiece, modulating the alternating stress / strain into an ACFM carrier signal via the inverse magnetostrictive effect; digitally orthogonally demodulating to extract the amplitude and phase; using a comprehensive sensitivity coefficient to invert the alternating strain; and spatially reconstructing to output the vibration displacement, velocity, acceleration, frequency, and mode shape. This invention achieves ACFM vibration measurement under rigid connection conditions, eliminates directional blind spots with a rotating magnetic field, ensures accuracy with closed-loop adjustment, and obtains spatial vibration distribution through array design.

Description

Technical Field

[0001] This invention relates to a mechanical vibration measurement system and method for ferromagnetic workpieces based on ACFM, belonging to the field of mechanical vibration measurement technology. Background Technology

[0002] Mechanical vibration measurement is an important tool for health monitoring, structural integrity assessment, and fault diagnosis of industrial equipment. Currently, commonly used vibration measurement methods mainly include contact vibration sensors, non-contact vibration measurement methods, and attempts at vibration measurement based on electromagnetic detection technology.

[0003] I. Contact Vibration Sensor Examples of such sensors include piezoelectric accelerometers, strain gauges, and magnetoelectric velocity sensors. These sensors typically need to be directly glued or rigidly mounted to the surface of the structure being measured, reflecting vibrations by sensing the structure's acceleration, velocity, or strain. Their advantages include mature technology and high sensitivity; however, they suffer from drawbacks such as complex wiring, limited channel count, difficulty in achieving large-area array measurements, and poor long-term reliability in chemical environments with high temperatures, corrosion, or strong electromagnetic interference. Furthermore, strain gauges can only measure localized strain and cannot directly obtain the spatial vibration distribution.

[0004] II. Non-contact vibration measurement methods These include laser Doppler vibration meters and eddy current displacement sensors. Laser vibration meters offer high accuracy and are non-contact, but their drawbacks include high cost, stringent requirements for the measurement environment (such as surface reflection conditions, smoke, and dust), and difficulty in simultaneously covering large areas and multiple points. Eddy current displacement sensors can measure the displacement of conductive material surfaces non-contactly, but they are essentially single-point measurements, and the signal is affected by both the lift-off distance and changes in the material's electromagnetic properties. In vibration measurements, they typically require an external fixed reference, making them unsuitable for rigid, servo-mounted applications.

[0005] For example, patent application publication number CN120507034A discloses a vibration signal detection device based on a magnetostrictive and resonance system. This device includes a worktable connecting chassis, a fixed disk, and a sensing component. The sensing component consists of a lower neodymium iron boron permanent magnet, an induction coil, an upper neodymium iron boron permanent magnet, and a magnetostrictive material. Based on the inverse magnetostrictive effect, it converts mechanical vibration into magnetic field changes, enabling vibration signal detection and fault diagnosis. Its shortcomings are: it uses a simple structure of a single-point permanent magnet + induction coil, lacks rotating magnetic field excitation, lacks an array design, cannot measure spatial vibration distribution, and does not involve the ACFM detection principle.

[0006] For example, patent application publication number CN103196989A discloses an ACFM crack detection system based on a rotating magnetic field. This system uses a rotating magnetic field signal excitation source to induce a uniformly rotating, uniformly strong electromagnetic field on the surface of the workpiece. The detection probe scans the surface of the workpiece to acquire the rotating magnetic field distortion signal caused by surface defects, thus achieving intelligent identification of cracks at different angles. Its shortcomings are: although it uses a rotating magnetic field for ACFM detection, it is only used for crack direction detection, does not involve vibration measurement, does not utilize the diamagnetic effect to sense stress / strain, and lacks closed-loop excitation control.

[0007] Furthermore, existing ACFM vibration measurements typically rely on changes in lift-off distance, i.e., using the change in magnetic field amplitude caused by the change in the distance between the probe and the workpiece surface to infer vibration displacement. However, this method has a drawback: it requires a relative displacement between the probe and the workpiece, necessitating maintaining a fixed reference position for the probe relative to the workpiece. In practical applications such as chemical containers, the probe is often rigidly mounted to the container wall and vibrates with the wall. In such cases, the lift-off distance remains constant, rendering the lift-off effect ineffective, and traditional ACFM vibration measurement methods are difficult to apply.

[0008] In summary, there is an urgent need for a novel vibration measurement system and method that can be rigidly mounted on the surface of a ferromagnetic workpiece, sense alternating stress / strain through changes in the electromagnetic properties of the material, and use arrayed sensors to measure spatial vibration distribution. Summary of the Invention

[0009] To address the technical problems of existing technologies, such as the reliance on scanning speed for quantitative defect measurement, the need for additional encoder configuration, the requirement for experimental calibration, and insufficient accuracy, the present invention aims to provide a mechanical vibration measurement system and method for ferromagnetic workpieces based on ACFM. Utilizing the ACFM detection principle under rotating magnetic field excitation, the mechanical vibration information reflected in the ACFM signal is obtained through the inverse magnetostrictive effect of the change in permeability of ferromagnetic materials under alternating stress / strain. This information is then acquired by an array of magnetic sensors and digitally demodulated. Alternating strain is obtained from the alternating magnetic field data, and finally, vibration data such as vibration displacement and mode shape are output through spatial reconstruction.

[0010] To achieve the above objectives, the present invention provides the following technical solution: The ACFM-based method for measuring mechanical vibration of ferromagnetic workpieces according to the present invention includes the following steps: S1. Establish a rotating alternating electromagnetic field: Apply a rotating alternating electromagnetic field excitation to the surface of the ferromagnetic workpiece; S2. Closed-loop adjustment of excitation magnetic field: Collect magnetic field data detected by each magnetic field sensor, determine the spatial uniformity of the rotating alternating electromagnetic field, and adjust the excitation parameters when a deviation is detected until the magnetic field uniformity meets the requirements. S3, ACFM signal acquisition: The magnetic field sensor rigidly mounted on the workpiece surface vibrates with the workpiece. The alternating stress / strain generated by the workpiece vibration causes a change in local permeability through the inverse magnetostrictive effect. The change in local permeability modulates the eddy current distribution induced by the rotating alternating electromagnetic field on the workpiece surface, causing the carrier signal detected by each magnetic field sensor to change. After signal conditioning, the discrete signal sequence is acquired by the data acquisition module. S4. Digital demodulation: Perform digital quadrature demodulation on the discrete signals of each channel to extract the amplitude and phase information of each channel signal; S5. Alternating strain inversion: The extracted amplitude information is processed to remove DC and drift to obtain the alternating component caused by vibration. The alternating strain at each measuring point is obtained by inversion using the pre-calibrated comprehensive sensitivity coefficient. S6. Vibration Data Output: Based on the spatial position of the sensor array and known structural parameters, the discrete strain data is spatially reconstructed using the relationship between strain and displacement in bending vibration. The vibration displacement of each measuring point is obtained by solving the problem, and the vibration velocity, acceleration, frequency and mode shape are further calculated to output the vibration data.

[0011] Furthermore, in step S1, the built-in DAC is controlled by the microcontroller system to generate two sinusoidal excitation signals DAC1 and DAC2 with adjustable amplitude and phase difference. After constant amplification by the excitation drive, they drive two sets of orthogonal coils respectively to induce a rotating alternating electromagnetic field on the surface of the ferromagnetic workpiece.

[0012] Furthermore, two orthogonal sinusoidal excitation signals are generated in the following way: a complete cycle of sine wave digital array is pre-stored in the Flash memory of the microcontroller system; a timer triggers DMA transfer to sequentially move the sample point data in the array to the data register of DAC1; the data source of DAC2 is the same as that of DAC1, but the starting read position is offset from the number of sample points corresponding to a quarter cycle in the array, thereby achieving a 90° phase difference between the two sinusoidal signals.

[0013] Furthermore, in step S2, an amplitude normalized variance threshold and a phase deviation threshold are set. When the amplitude normalized variance or phase deviation of each magnetic field sensor exceeds the corresponding threshold, a proportional control strategy is used to adjust the output amplitude of the corresponding DAC channel or the relative phase offset of the two signals until the magnetic field uniformity meets the requirements.

[0014] Further, in step S4, the microcontroller system synchronously or polls the signals of each channel, generates in-phase and quadrature reference signals with the excitation frequency as the reference frequency, multiplies them by the sampling sequences of each channel respectively, and then filters out the second harmonic component through a digital low-pass filter to obtain the in-phase component. and orthogonal components ,Depend on and Calculate the amplitude of each channel. and phase time .

[0015] Furthermore, in step S5, for ferromagnetic materials, the diamagnetic effect gives an approximately linear relationship between the change in magnetic permeability and strain: ; in, This represents the change in permeability of a ferromagnetic material under alternating stress / strain. The permeability of ferromagnetic materials, This represents the relative change in magnetic permeability. This is the stress sensitivity coefficient. For elastic modulus, For alternating strain; The relative change in ACFM signal amplitude is proportional to the change in permeability: ; in, The alternating amplitude component of the ACFM detection signal caused by vibration. This is the sensitivity factor of the system to changes in permeability. This is the static operating point amplitude; By combining the approximately linear relationship between permeability change and strain, and the relationship between the relative change in ACFM signal amplitude and permeability change, a direct relationship between the relative change in amplitude and alternating strain is obtained: ; Among them, the comprehensive calibration coefficient , is a dimensionless constant, which is obtained in one calibration through a standard specimen vibration table experiment; Alternating strain is obtained by inversion from alternating magnetic field data: ; in, The average value of the static operating point amplitude; the DC removal and drift removal processes employ high-pass filtering or moving average.

[0016] Furthermore, in step S6, it is assumed that the workpiece wall is a thin plate structure, bending vibration is dominant, and surface strain and lateral displacement are... The relationship is: ; in, For wall thickness, For the Laplace operator; The spatial reconstruction includes the following two methods: Method 1: When the measured area is far from structural discontinuities, the finite difference method is used for internal nodes. The second derivative is approximated by the central difference: ; in, , The arrays are respectively in direction and Spacing in direction, For nodes Vibration displacement at the location, , They are nodes exist Vibration displacement of adjacent nodes in the direction, , They are nodes exist Vibration displacement of adjacent nodes in the direction; node The relationship between strain and displacement is as follows: ; Rewrite the above equation in matrix form. , For discrete Laplace matrices, magnetic sensor array The displacement vector is formed by the vibration displacements of each node; the displacement is solved by combining the boundary conditions. Furthermore, a symmetrical extension process is employed to treat the sensor array as the central part of a larger grid, and the strain values ​​of virtual nodes are extrapolated to ensure the well-posedness of the equation system. Method 2: When the area under test is close to a structural discontinuity, the modal superposition method is used: the first few vibration modes of the area under test are obtained in advance through finite element modal analysis or experimental hammer test. and their corresponding strain mode shapes ,in, The modal order index is used. For the first First mode at position The mode shape function at that point, For the first Strain mode shapes of the first mode; solve for the modal coordinates by projecting the measured strain vector onto the strain mode shapes. ,in, For the first First mode with time Reconstructing the displacement field from the changing modal coordinates: .

[0017] Furthermore, in step S3, the influence of static lift-off distance variation on the measurement results is eliminated through amplitude self-normalization: the amplitude at the static operating point of each channel is used as the standard. For alternating components Normalization yields the relative change. ,in, Channel number, For the first The amplitude alternation component of the channel, For the first The average value of the channel's static operating point amplitude.

[0018] The ACFM-based mechanical vibration measurement system for ferromagnetic workpieces described in this invention, applied to the ACFM-based mechanical vibration measurement method for ferromagnetic workpieces, includes: The microcontroller system contains a timer, waveform array memory, two DAC channels, an MCU processing core, and a built-in multi-channel ADC. It is used to generate two sinusoidal excitation raw signals DAC1 and DAC2 with adjustable amplitude and phase difference, and to receive the sampled signals from each channel for digital demodulation. The excitation drive is electrically connected to the two DAC channels of the microcontroller system and includes a first power amplifier and a second power amplifier, which are used to amplify the two sinusoidal excitation signals with constant gain respectively. The excitation coil group, electrically connected to the excitation drive, includes a first coil group consisting of a first coil and a second coil connected in series, and a second coil group consisting of a third coil and a fourth coil connected in series, used to induce a rotating alternating electromagnetic field on the surface of a ferromagnetic workpiece. A magnetic sensor array, rigidly mounted above the excitation coil group and vibrating with the workpiece, includes multiple magnetic sensors arranged in an array. Each magnetic sensor detects a magnetic field component parallel to the workpiece surface along the direction of the main excitation current and a magnetic field component perpendicular to the workpiece surface. The signal conditioning module, electrically connected to the magnetic sensor array, includes a bias circuit and an instrumentation amplifier. It amplifies and DC biases the weak differential signal output by the magnetic sensor array before inputting it into the built-in multi-channel ADC of the microcontroller system.

[0019] Furthermore, the system also includes an outer housing, with the excitation coil group placed inside the housing at the bottom layer and in close contact with the surface of the ferromagnetic workpiece being measured, the magnetic sensor array placed on the upper layer of the excitation coil group, the signal conditioning module placed on the upper layer of the magnetic sensor array, and the microcontroller system placed on the top layer; a shielding layer and a heat insulation layer are provided between each layer to suppress the direct coupling of the excitation field to the magnetic sensor array and reduce temperature drift.

[0020] Compared with existing technologies, the ACFM-based mechanical vibration measurement system and method for ferromagnetic workpieces of the present invention exhibits the following beneficial effects in terms of technical performance and practical application: 1. ACFM vibration measurement under rigid connection conditions has been achieved. This invention abandons the traditional approach of relying on changes in lift-off distance in ACFM vibration measurement, and instead utilizes the inherent inverse magnetostriction effect of ferromagnetic materials to directly modulate the alternating stress / strain caused by vibration into the ACFM signal. The probe can be rigidly mounted on the workpiece surface and vibrates with the workpiece, without the need for an external fixed reference datum, solving the problem that probes can only be mounted in motion in applications such as chemical containers.

[0021] 2. Rotating magnetic field excitation eliminates the directional blind zone. This invention uses four-region orthogonal coils to generate a rotating alternating magnetic field. Compared with the unidirectional excitation of traditional ACFM, it can maintain high sensitivity to alternating stress in any direction, avoiding the measurement blind zone problem caused by the mismatch between the stress direction and the excitation field direction, and significantly improving the reliability and consistency of vibration measurement.

[0022] 3. Closed-loop excitation adjustment ensures measurement accuracy. This invention monitors sensor data in the x-direction of the sensor array in real time and adjusts the excitation coil drive parameters in a closed loop to ensure the spatial uniformity of the rotating magnetic field within the sensor area. This adaptive mechanism effectively compensates for magnetic field distortion caused by factors such as coil manufacturing tolerances, changes in workpiece surface curvature, and local material permeability inhomogeneity, significantly improving measurement repeatability and accuracy.

[0023] 4. A 10kHz excitation frequency enables low-cost, high-precision digital demodulation. This invention limits the excitation frequency to 10kHz, allowing the microcontroller's built-in ADC to directly sample multiple signals synchronously or poll at a sampling rate of 100kSPS, eliminating the need for expensive external high-speed ADCs or dedicated analog lock-in amplifier chips. Digital quadrature demodulation is completed within the microcontroller, ensuring the extraction accuracy of weak vibration modulation signals while significantly reducing system hardware costs and complexity, making it suitable for mass deployment in industrial settings.

[0024] 5. Arrayed design enables acquisition of spatial vibration distribution. This invention employs a 3×3 dual-axis magnetic sensor array, capable of simultaneously acquiring magnetic field change information at multiple spatial points within the measurement area. Combined with spatial difference and numerical reconstruction algorithms, vibration displacement field and spatial mode shapes can be obtained from discrete strain data, overcoming the limitation of traditional single-point vibration sensors that can only acquire single-point time histories, and providing richer data support for structural modal analysis and damage localization.

[0025] 6. Strong environmental adaptability, suitable for harsh chemical working conditions. The functional layers of the probe of this invention adopt a stacked structure, and electromagnetic shielding and heat insulation layers can be set between the layers to effectively suppress direct excitation-sensing coupling and temperature drift. The probe as a whole adopts a rigid package, which can adapt to the high temperature, corrosion, vibration and explosion-proof requirements common in chemical containers, and its long-term reliability is better than that of traditional adhesive strain gauges and accelerometers.

[0026] 7. High system integration, facilitating engineering applications. This invention integrates excitation, sensing, signal conditioning, and data processing into a single housing, resulting in a compact structure and convenient installation. Measurement can begin simply by placing the bottom surface of the housing tightly against the surface of the ferromagnetic workpiece being measured, eliminating the need for complex optical path adjustments or external reference setups. This makes it particularly suitable for long-term online vibration monitoring of specific locations on chemical containers (such as near welds, nozzle roots, and locally thinned areas). Attached Figure Description

[0027] Figure 1 This is a flowchart of the measurement method of the present invention; Figure 2 This is a block diagram of the overall system composition of the present invention; Figure 3 This is a schematic cross-sectional view of the mechanical housing structure of the present invention; Figure 4 This is a schematic diagram of the four-region coil excitation layout of the present invention; Figure 5 This is a connection diagram of the first and third coils with the DAC output channel in this invention; Figure 6 This is a connection diagram of the second and fourth coils with the DAC output channel in this invention; In the diagram: 1. Outer shell; 2. Shielding layer; 3. Heat insulation layer; 4. Excitation drive; 5. Magnetic sensor array; 6. Microcontroller system; 7. First coil; 8. Second coil; 9. Third coil; 10. Fourth coil. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0029] Example 1: The overall process of the ACFM-based mechanical vibration measurement method for ferromagnetic workpieces described in this invention is as follows: Figure 1 As shown, it includes the following steps: S1. Establishing a rotating alternating electromagnetic field: Apply a rotating alternating electromagnetic field excitation to the surface of the ferromagnetic workpiece. Four-region coils are arranged on the PCB board, forming two sets of orthogonal coil groups in pairs. For example... Figures 4-6As shown, the first coil 7 and the second coil 8 are connected in series to form the first coil group, and the third coil 9 and the fourth coil 10 are connected in series to form the second coil group. The two coil groups are driven by the excitation driver 4 (i.e., Class AB power amplifier) ​​through two DAC output channels, respectively, to induce a rotating alternating electromagnetic field on the surface of the ferromagnetic workpiece. The microcontroller system 6 controls the built-in DAC to generate two 10kHz sine wave signals through DMA + waveform array + timer. A complete cycle of sine wave digital array (256 sampling points) is pre-stored in the flash memory of the microcontroller system 6. DAC1 sequentially moves the array data, and DAC2 starts reading from a position offset by 64 sampling points (one-quarter of a cycle). The two signals are constantly amplified by the excitation driver 4 and then drive the two sets of quadrature coils respectively.

[0030] S2. Excitation Magnetic Field Closed-Loop Adjustment: Collect magnetic field data detected by each magnetic field sensor in the magnetic sensor array 5, determine the spatial uniformity of the rotating magnetic field, and adjust the excitation parameters when a deviation is detected until the magnetic field uniformity meets the requirements. Figure 1 As shown, this step first reads the x-direction signal from the 3×3 sensor, calculates the amplitude and phase at each point, compares it with the ideal uniform field, and then adjusts the DAC amplitude or phase. Next, a judgment node is used to determine if the magnetic field uniformity meets the requirements. If the magnetic field uniformity does not meet the requirements, the process returns to the closed-loop adjustment loop to readjust the DAC amplitude / phase, forming a closed-loop feedback loop. If the magnetic field uniformity meets the requirements, the process proceeds to the next step. A normalized variance threshold for amplitude and a phase deviation threshold are set. When these thresholds are exceeded, a proportional control strategy is used to adjust the output amplitude of the DAC channel or the relative phase shift between the two signals.

[0031] S3, ACFM Signal Acquisition: The magnetic sensor array 5, rigidly mounted on the workpiece surface, vibrates along with the workpiece. The alternating stress / strain generated by the workpiece vibration causes a change in local permeability through the inverse magnetostrictive effect. This change in permeability modulates the eddy current distribution induced by the rotating magnetic field on the workpiece surface, causing a change in the carrier signal detected by each sensor. The output signals of each sensor are amplified and DC biased by the analog signal processing module in the microcontroller system 6, and then acquired by the data acquisition module to obtain a discrete signal sequence.

[0032] S4. Digital Demodulation: Perform digital quadrature demodulation on the discrete signals of each channel to extract the amplitude and phase information of each channel signal. In this step, the data from each ADC channel is multiplied by a reference signal and then low-pass filtered to obtain the in-phase component. and orthogonal components ,Depend on and Calculate the amplitude time series of each channel and phase time series The microcontroller system generates in-phase and quadrature reference signals at a reference frequency of 10kHz.

[0033] S5. Alternating Strain Inversion: The extracted amplitude information is processed to remove DC and drift, yielding the alternating component caused by vibration. The alternating strain at each measuring point is then inverted using a pre-calibrated comprehensive sensitivity coefficient. This step extracts the amplitude sequence. alternating components in According to the calibration coefficient Calculate the strain at each measuring point Using the dimagnetostriction effect formula Relationship with ACFM signal The comprehensive calibration coefficients were derived. .

[0034] S6. Vibration Data Output: Based on the spatial positions of each sensor in the magnetic sensor array 5 and the known structural parameters, the discrete strain data is spatially reconstructed using the relationship between strain and displacement in bending vibration. The vibration displacement at each measuring point is then calculated, and the vibration velocity, acceleration, frequency, and mode shape are further calculated, outputting the vibration data. In this step, the displacement field is reconstructed from the discrete strain data through spatial difference / modal superposition. The vibration parameters are output, and the process ends. Spatial reconstruction is performed using either the finite difference method (Method 1) or the modal superposition method (Method 2).

[0035] Example 2: Based on Example 1, this example corresponds to step S5 of the above method, which involves "using a pre-calibrated comprehensive sensitivity coefficient". The alternating strain at each measuring point is obtained through inversion, which is used to establish the relative change in the amplitude of the ACFM signal. With alternating strain The quantitative relationship between them.

[0036] 1. Physical meaning and dimensional derivation of the comprehensive calibration coefficient K Comprehensive calibration coefficient It consists of three parameters: (1) Stress sensitivity coefficient (piezomagnetic coefficient) The relative rate of change of magnetic permeability of ferromagnetic materials under unit stress, originating from the dimagnetic stretching effect. Dimensions are... For low-carbon steel and low-alloy steel (such as Q345R). The typical range is The Q345R material used in this embodiment has been measured. .

[0037] (2) : Elastic modulus (Young's modulus) The ratio of stress to strain in a material within its elastic range. Dimensions are MPa. Carbon steel. MPa; 304 stainless steel MPa.

[0038] (3) Permeability-Magnetic Sensitivity Factor Characterizing the relative change in magnetic permeability of materials Causes relative changes in the amplitude of the ACFM detection signal The conversion efficiency depends on the probe geometry, excitation frequency, skin depth, and the operating point of the magnetic sensor; it is a dimensionless system parameter. For the 10kHz excitation, TMR sensor, and rotating magnetic field configuration used in this invention, The measured value is approximately 0.15~0.25. In this embodiment, the value was obtained through fitting. .

[0039] (4) Dimensionless derivation Substitute the dimensions of each factor: , , ,but .therefore, This is a dimensionless constant. Substituting the above typical values ​​into: This is consistent with the subsequent calibration results.

[0040] 2. Experimental Apparatus and Specimens This embodiment uses the measurement system described in this invention for calibration experiments. The specimen is made of Q345R carbon steel plate, the same material as the chemical container being tested, with dimensions of 200mm × 50mm × 8mm. The surface is polished smooth and free of oxide scale. The excitation and sensing adopt the same probe structure as the system of this invention. The excitation frequency is f0 = 10kHz. The excitation signal is generated by the DAC in the single-chip microcomputer system 6 and amplified by a Class AB power amplifier to establish a rotating magnetic field on the surface of the specimen.

[0041] A high-precision resistance strain gauge (model BF350-3AA, sensitivity coefficient 2.0) is attached to the back of the specimen directly below the center of the probe. The strain gauge leads are connected to a dynamic strain gauge (such as DH3817) to output the known reference strain. .

[0042] The vibration table is an electromagnetic vibration table (such as LDSV406), which can generate sinusoidal vibrations with adjustable frequency and amplitude. The specimen is rigidly fixed on the vibration table, and the probe is tightly attached to the surface of the specimen directly above the strain gauge by bolts, ensuring that the lift-off distance is zero (or controlled at a constant value of ≤0.1mm).

[0043] 3. Calibration Operation Steps Step 1: According to Figure 1 The method flow shown is as follows: start the system, establish a rotating magnetic field (step S1), and perform closed-loop adjustment (step S2) to ensure that the magnetic field uniformity in the sensor area meets the requirement of deviation ≤5%. Preheat the system for 30 minutes to allow the temperature to stabilize.

[0044] Step 2: Set the frequency generated by the vibration table The Hz sinusoidal vibration was adjusted so that the strain amplitude on the specimen surface was approximately [value missing]. During the vibration process, the following were recorded simultaneously: Reference strain signal output by dynamic strain gauge Sampling rate f s =1kHz; Relative change in ACFM demodulation amplitude of the probe center channel .

[0045] Among them, the single-chip microcomputer system 6 uses f s The original signal from this channel sensor is sampled at 100kSPS and digital quadrature demodulation is performed (corresponding to step S4) to obtain the amplitude sequence. Take the average value of the data from the first 10 seconds as the static amplitude average value. ; Calculate alternating components ; Obtain the relative change sequence The synchronous acquisition lasted for 20 seconds and included at least 2000 vibration cycles.

[0046] Step 3: Based on the linear relationship Using records and Perform least squares fitting: ; In the formula: q(t) represents the normalized amplitude change obtained by digital quadrature demodulation of the magnetic sensor over continuous time, t is time, and q(n) is the value of its discrete sampling points (i.e., ); ε(t) represents the reference strain applied at the corresponding time, and the discrete sequence is denoted as That is, the magnetic sensor array ε(t)=q(t) / K is used for subsequent alternating strain inversion at each measuring point. Calculations yielded .

[0047] Step 4: Change the vibration table frequency to 50Hz, 200Hz, and 500Hz, and repeat steps 2-3 to obtain the K values ​​at different frequencies as shown in Table 1 below: Table 1. Comprehensive calibration coefficient K at different vibration frequencies

[0048] It can be seen that K varies by less than 10% in the range of 50~500Hz, and is approximately independent of frequency. The average K=0.33 (or fitted as a function of frequency) is written into the calibration parameter table of the microcontroller system.

[0049] Step 5: Temperature compensation verification. The calibration was repeated at room temperature (25℃) and 50℃, respectively, and the K values ​​were 0.33 and 0.32, respectively.

[0050] 4. Results Calibration yields comprehensive coefficients (Corresponding frequency 50~500Hz, temperature 25~50℃). In subsequent actual measurements, the microcontroller system 6 used this K value, through the formula... Calculate the alternating strain at each measuring point. Among them, Let be the alternating strain at the i-th measuring point at time t. For the first The amplitude alternating component of the channel, magnetic sensor array For the first The average value of the channel's static operating point amplitude.

[0051] Example 3: This embodiment corresponds to all steps S1 to S6 in the above method, and specifically implements the excitation signal generation method in step S1, the closed-loop adjustment strategy in step S2, the digital demodulation algorithm in step S4, the alternating strain inversion in step S5, and the finite difference method displacement reconstruction in step S6. A simply supported rectangular carbon steel plate is used as the test object to verify the system's ability to measure bending vibration.

[0052] 1. The object under test and the installation The specimen was a Q345R steel plate, measuring 800mm × 500mm × 6mm, simply supported on all four sides, and placed horizontally. A small electromagnetic vibrator was installed below the center of the plate, and a frequency f was applied. v =120Hz sinusoidal excitation. The probe of this invention (3×3 array, array spacing dx=dy=20mm) is rigidly mounted in the upper left area of ​​the board surface, with the center about 200mm from the center of the board. The bottom of the probe is in close contact with the board surface without gaps.

[0053] 2. Measurement Procedure Rotating magnetic field establishment: The timer in the microcontroller system 6 triggers the DMA at 2.56MHz, and the DAC outputs two 10kHz sinusoidal signals with a 90° phase difference (amplitude). ), driven by excitation 4 (Class AB power amplifier, gain Amplified drive such as Figure 4 The four-region coil shown generates a rotating magnetic field on the surface of the plate.

[0054] The first coil 7 and the second coil 8 form the first coil group, driven by the DAC1 channel via excitation driver 4; the third coil 9 and the fourth coil 10 form the second coil group, driven by the DAC2 channel via excitation driver 4. DAC1 output. DAC2 output ,in, The excitation frequency is the frequency of the sinusoidal signal output by the DAC in the microcontroller system 6. , These are the sinusoidal excitation signals output from channels DAC1 and DAC2, respectively. After being amplified by the power amplifier, the two signals are output as follows: , ,in, , These are the driving voltages output from the excitation driver 4 to the two sets of coils, driving the two sets of orthogonal coils.

[0055] Under ideal uniform magnetic field conditions, the position of the 3×3 sensor within the plane is... The magnetic field component in the x-direction at that location is ,in, As a reference for magnetic field amplitude, A fixed phase determined by spatial location.

[0056] Excitation magnetic field closed-loop regulation: The ADC synchronously acquires signals from 9 x-direction sensors at 100kSPS, and obtains the x-direction magnetic field amplitude at each point through digital quadrature demodulation. and phase Calculate the average amplitude. and phase average Calculate the normalized amplitude deviation for each channel. and phase deviation Initial measurements showed a maximum amplitude deviation of 12%. or At that time, a proportional control strategy is used to adjust the DAC output: , .in, , These are the DAC output amplitudes before and after adjustment, respectively. , These represent the relative phase shifts between the two signals before and after adjustment. , These are the proportional control coefficients for amplitude and phase, respectively.

[0057] like Figure 1As shown, the magnetic field uniformity is judged by the judgment node to determine whether it meets the requirements. If it does not meet the requirements, it returns to the closed-loop adjustment loop for readjustment. After closed-loop adjustment (adjusting the output amplitude ratio of the two DACs), the maximum deviation is reduced to 3% after 10 iterations, which meets the threshold requirement, and then proceeds to the next step.

[0058] ACFM signal acquisition: Vibration continues, and the differential voltage signals output by each sensor in the magnetic sensor array 5 are recorded. ,in, For the first The differential voltage signal output by each sensor. The sensor sensitivity (V / T) is given. For the first The magnetic field strength at the location of each sensor. The analog signal processing module in the microcontroller system amplifies the signal using an instrumentation amplifier and applies DC bias before outputting it to the ADC. ,in, For the first The signal input to the ADC after amplification and biasing is from the channel. For instrumentation amplifier gain, This is the DC bias voltage. The ADC discretizes the continuous signal from each channel: ,in, For the first Discrete sampling sequence of the channel, Sampling time, The sampling rate is [value]. Sensor signals from each channel (x-direction and z-direction) are acquired synchronously for 10 seconds continuously.

[0059] Digital quadrature demodulation: Digital quadrature demodulation is performed on each of the 18 signals. First, a digital reference signal with the same frequency as the excitation signal is generated in the microcontroller system 6. , .in, For in-phase reference signal, This is an orthogonal reference signal. Then, multiplication mixing is performed: , .in, , These are the signals after mixing with in-phase and quadrature reference signals, respectively. The baseband components are obtained after digital low-pass filtering (FIR low-pass filter cutoff frequency 500Hz, order 128): , .in, This is a digital low-pass filter operation; multiplying by 2 is to compensate for the 1 / 2 amplitude loss introduced by mixing. , The first The in-phase and quadrature components of the channel. (By...) and Calculate the amplitude time series and phase time series .in, For the first The amplitude time series of the channel, For the first Phase time series of the channel.

[0060] Alternating strain inversion: amplitude sequence for each channel The average amplitude at the static operating point is obtained by taking the average value over 10 seconds. .in, This represents the total number of sampling points averaged over a long period. High-pass filtering removes slow drift, yielding the alternating component. High-pass filtering (cutoff 0.5Hz) removes slow drift, yielding the alternating component. Using calibration coefficients Calculate the strain at each measuring point: ; in, For the i-th measurement point at time t, Alternating strain, This represents the alternating amplitude component of the i-th channel, i.e., the alternating portion of the i-th channel's amplitude sequence after removing DC and slow drift. The average value of the static operating point amplitude of the i-th channel is obtained by long-term averaging; Vibration displacement reconstruction: using strain data from a 3×3 array, a discrete Laplace matrix is ​​constructed using central difference. Given the plate thickness mm, with boundary conditions set to zero displacement at all four boundary points. The array is treated as part of a larger plate region using symmetric extension, and reconstructed using modal superposition.

[0061] Specifically, the first three bending modes of the plate are known: ; in, The modal order index is used. , These are the length and width of the plate, respectively. The relationship between strain and displacement is as follows: Therefore, the strain mode shape is The strain time series of the nine measuring points were projected onto... The coordinates of each modality are obtained above. Then reconstruct the displacement .

[0062] Output results: Displacement time history: The displacement of the array center point (position of sensor 5) was reconstructed. Its amplitude is approximately 4.2 μm and its frequency is 120 Hz.

[0063] Velocity: Velocity is obtained by numerical differentiation. The velocity amplitude is approximately 3.2 × 10⁻⁶. 3 m / s.

[0064] Acceleration: Acceleration is obtained by numerical differentiation. The acceleration amplitude is approximately 2.4 m / s².

[0065] Mode shape identification: FFT was performed on the displacement of 9 measuring points to extract the spatial distribution of the 120Hz component, which is in good agreement with the theoretical sinusoidal mode shape, with a relative error of less than 8%.

[0066] This embodiment verifies that the system can invert the displacement, velocity, acceleration and mode shape information of bending vibration from the rigidly installed ACFM array signal without relying on an external reference datum, thus meeting the engineering monitoring requirements.

[0067] Example 4: This embodiment corresponds to steps S1 to S6 in the method of Embodiment 1, and the displacement reconstruction using the modal superposition method described in step S6, Method 2. It demonstrates the application of the system in a typical part of a real chemical container (the root of the nozzle) and its data output.

[0068] 1. The object under test and its operating conditions The container is a vertical carbon steel storage tank (Q345R), with a wall thickness of 10mm, a diameter of 2m, a height of 5m, and a normal operating pressure of 0.5MPa. The medium is water. There is a DN200 connecting pipe 1.5m from the bottom of the tank wall. The vibration source is random vibration in the root area of ​​the connecting pipe caused by fluid pulsation in the pipeline and pump operation, with the main frequency band concentrated in the range of 80~300Hz.

[0069] The system of the present invention (3×3 array probes, 15mm spacing) is rigidly welded or magnetically fixed to the tank wall near the root of the nozzle, with the center of the array 20mm away from the weld of the nozzle and the direction along the circumference and axial direction of the tank wall.

[0070] 2. Measurement Procedure like Figure 1 The method flow shown is implemented step by step as follows: S1-S2: Establish a rotating magnetic field and complete closed-loop regulation. Due to the curvature of the tank wall, the initial magnetic field non-uniformity reaches 20%, which is reduced to 4% after closed-loop regulation.

[0071] S3-S4: Continuously acquire 18 channels of signals for 10 minutes (sampling rate 100kSPS), perform real-time digital quadrature demodulation, and extract the amplitude and phase of each channel. Output a set of amplitude data every 1 second.

[0072] S5: Calculate the static mean for the amplitude data for each second. and alternating components The strain time series was converted using a calibration K=0.33 (temperature compensation already performed).

[0073] S6: Due to the complex stress distribution at the nozzle root, simple differential reconstruction is not used; instead, modal superposition is employed. The first six vibration modes of the local region at the nozzle root are obtained in advance through finite element modal analysis. ( ), and calculate its corresponding strain mode shape. The strain vectors from the nine measuring points are projected onto the strain mode shape, and the modal coordinates are solved. Then reconstruct the displacement field: ; The selection of the modal truncation order is based on the principle that the modal participation mass of the truncated mode within the measurement frequency band (e.g., 0~500Hz) reaches more than 90% of the total mass. For the local region in this embodiment, the first 6 modes are sufficient. Error estimation is achieved by calculating the residuals. ; in, The measured strain vector is used, and the reconstruction accuracy is considered to meet engineering requirements when Res < 5%. In this embodiment, the residual is 3.2%, verifying the rationality of M=6.

[0074] 3. Results Output Vibration displacement spectrum: FFT was performed on the displacement of the center measuring point, and three main frequency components were identified: 112Hz (amplitude 3.8μm), 178Hz (amplitude 2.1μm), and 245Hz (amplitude 1.2μm).

[0075] The RMS values ​​of displacement at each measuring point are shown in Table 2 below: Table 2. RMS values ​​of vibration parameters at various measuring points at the pipe root.

[0076] Mode shape identification: The spatial distribution of displacement at each measuring point corresponding to the 112Hz component was extracted and found to be consistent with the local mode shape at this frequency predicted by the finite element method, with a correlation coefficient of 0.92, which verified the effectiveness of the modal superposition reconstruction.

[0077] Anomaly detection: Within 10 minutes of continuous monitoring, the strain and displacement amplitudes of each channel remained stable, without any sudden increases or frequency drift, indicating that the vibration level at the root of the connector was within the normal range.

[0078] Example 5: This embodiment corresponds to the ACFM signal acquisition described in step S3 of the method in embodiment 1, and the amplitude self-normalization processing in step S3 to eliminate the influence of static lift-off distance variation on the measurement results.

[0079] 1. Experimental setup The object under test is a Q345R carbon steel plate, measuring 500mm × 300mm × 6mm, simply supported on all four sides. An electromagnetic vibrator applies a frequency at the center of the plate. A stable sinusoidal excitation with a frequency of 10 kHz and amplitude induces stable bending vibrations on the plate surface. The probe uses the same 3×3 biaxial sensor probe as in this invention, with an excitation frequency of 10 kHz and a rotating magnetic field. The lift-off distance between the probe bottom and the plate surface is adjusted using non-magnetic pads (PTFE sheets) of varying thicknesses. Take respectively mm (direct contact), 0.5mm, 1.0mm, 2.0mm. A resistance strain gauge is attached to the back of the plate directly below the probe to output the reference strain. The strain amplitude is constant. .

[0080] 2. Experimental Procedure At each liftoff distance, rotating magnetic field excitation setup and closed-loop adjustment (S1, S2) were performed, and the average amplitude of the static operating point of each channel was recorded. .

[0081] The signals from each channel were continuously acquired for 10 seconds and then digitally quadrature demodulated to obtain the signal. Calculate alternating components and relative changes .in, This represents the relative change in amplitude of the i-th channel at time t, i.e., the relative change ratio of the ACFM signal amplitude of the i-th channel. Let be the alternating amplitude component of the i-th channel at time t, which is the alternating part of the i-th channel amplitude sequence after removing DC and slow drift.

[0082] Using calibration coefficients Inversion strain The data from the central measuring point (channel 5) were compared with the reference strain.

[0083] 3. The experimental results are shown in Table 3 below: Table 3 Measurement results at different lift-off distances

[0084] 4. Analysis As the lift-off distance increases from 0 to 2mm, the static operating point amplitude... The voltage decayed exponentially from approximately 2.35V to 0.86V, a decrease of about 63%. However, the relative change after self-normalization... The amplitude remained basically constant (0.97×10). -6 ~1.01×10 -6 The fluctuation is within ±2%. The strain amplitude obtained from the inversion is similar to the reference value. The relative errors were all within ±2%, indicating that variations in lift-off distance were effectively eliminated through static amplitude normalization, and the measurement results were independent of lift-off distance. Conclusion: This invention... The self-normalization process can maintain strain measurement accuracy within the lift-off distance range of 0~2mm, proving that the solution is robust to changes in installation gap and surface coating thickness.

[0085] Example 6: This embodiment corresponds to the technical feature described in step S5 of the above method, where the comprehensive sensitivity coefficient K is approximately constant within the strain amplitude range of 0~50με and the linearity deviation is less than 3%, as well as the mathematical well-stability description of the symmetric extension processing in step S6.

[0086] 1. Linear range of strain amplitude Maintaining a temperature of 25℃ and an excitation frequency of 10kHz, sinusoidal vibrations of different amplitudes were applied to the Q345R specimen, and measurements were taken. The relationship between the strain amplitude and the strain amplitude is shown in Table 4 below: Table 4. Strain amplitudes at different values linearity deviation

[0087] in, Let be the strain amplitude of the applied sinusoidal vibration. Conclusion: In the strain amplitude range of 0~50με (i.e., Within the specified range, a linearity deviation of less than 3% can be considered a linear operating region. When the strain exceeds 100 με, the change in permeability tends to saturate, and the nonlinear deviation exceeds 5%. Therefore, this invention defines the effective linear measurement range as 0~50 με.

[0088] 2. Temperature range Keeping the strain amplitude constant at 2με, the changes in the ambient temperature and the measured changes in the K value are shown in Table 5 below: Table 5. Comprehensive calibration coefficient K at different temperatures

[0089] Conclusion: Within the temperature range of -20 to 80°C, the maximum relative deviation of the K value from the 25°C reference (K=0.33) is approximately... 12.1% At 20°C, the deviation does not exceed ±10% within the range of 0~80°C. After linear correction by the built-in temperature sensor of the single-chip microcomputer system, the measurement error can be controlled within ±5%. When the temperature exceeds 100°C, the magnetic properties of the material degrade significantly, and it is recommended to use a high-temperature dedicated probe. Therefore, the effective operating temperature range of this invention is defined as -20~80°C.

[0090] Mathematical well-posedness explanation of reconstructing the displacement field using a 3.3×3 array For thin plate bending problems, surface strain and lateral displacement satisfy On a 3×3 regular mesh, the central difference approximation of the second derivative is applied to the internal nodes, resulting in 9 linear equations with the 9 nodal displacements as the unknowns. However, the Laplace operator of the boundary nodes located at the edge of the array needs to reference the nodes outside the array, and the equation system cannot be closed by relying solely on the internal data of the 3×3 array. This invention expands the number of equations by treating the 3×3 array as the central part of a larger grid through symmetric extension, or by using the modal superposition method of method two to project the measured strain onto the strain mode shape to solve the modal coordinates, thereby ensuring that the equation system is well-posed and the solution is stable.

[0091] The selection criteria for boundary conditions are as follows: For areas far from structural discontinuities (such as the middle of a cylindrical section), local regions can be considered approximately simply supported around the perimeter, meaning the boundary node displacements are minimal. However, the corners are free. To ensure well-posedness, a symmetric extension is used: the 3×3 array is considered as the central part of a larger 5×5 or 7×7 uniform grid. Utilizing structural symmetry and the smoothness of strain distribution, the strain values ​​of the virtual nodes are extrapolated, thus increasing the number of internal equations. A ring of virtual nodes (16 virtual points in total) is added outside the 3×3 grid, and their strain values ​​are obtained by mapping the strain values ​​of the inner nodes according to symmetry. For example, for modes symmetric about the x and y axes, the strain of the virtual nodes satisfies... .in, , Let the strains at a pair of virtual nodes and the strain at a pair of actual nodes be symmetric about the coordinate axes, respectively. In a symmetric extension, the two are equal. This expands the number of equations to 25 (5×5 grid), with 25 unknowns. Combined with boundary conditions (the outermost ring is fixed or simply supported), the system of equations is well-posed and the solution is stable.

[0092] Near structural discontinuities (such as pipe roots and welds): the modal superposition method is used to obtain the first few vibration modes of the local area in advance through a finite element model. and their corresponding strain mode shapes The measured strain vector Solving for modal coordinates by projecting onto strain mode shapes Then reconstruct the displacement.

[0093] Example 7: The ACFM-based mechanical vibration measurement system for ferromagnetic workpieces of this invention comprises: a microcontroller system 6, internally including a timer, waveform array memory, two DAC channels, an MCU processing core, and a built-in multi-channel ADC, used to generate two sinusoidal excitation raw signals DAC1 and DAC2 with adjustable amplitude and phase difference, and to receive the sampled signals from each channel for digital demodulation; an excitation driver 4, electrically connected to the two DAC channels of the microcontroller system 6, including a first power amplifier and a second power amplifier, used to amplify the two sinusoidal excitation signals with constant gain respectively; and an excitation coil group, electrically connected to the excitation driver 4, including a first coil 7 and a second coil 8 connected in series to form a... A second coil group, consisting of a first coil group, a third coil 9, and a fourth coil 10 connected in series, is used to induce a rotating alternating electromagnetic field on the surface of a ferromagnetic workpiece. A magnetic sensor array 5, rigidly mounted above the excitation coil group and vibrating with the workpiece, includes multiple biaxial magnetic sensors arranged in an array. Each magnetic sensor detects a magnetic field component parallel to the workpiece surface along the direction of the main excitation current and a magnetic field component perpendicular to the workpiece surface. A signal conditioning module, electrically connected to the magnetic sensor array 5, includes a bias circuit and an instrumentation amplifier. It amplifies and DC biases the weak differential signal output by the magnetic sensor array 5 before inputting it into the built-in multi-channel ADC of the microcontroller system 6.

[0094] The specific system architecture is as follows: Figure 2 As shown, the top layer is the microcontroller system 6, which internally includes a timer, waveform array memory, two DAC channels, an MCU processing core, and a built-in multi-channel ADC. The timer outputs timing signals to the waveform array memory, and the waveform array memory outputs signals to the two DAC channels. Simultaneously, the MCU processing core can write back to the waveform array memory, and the signals acquired by the ADC are input to the MCU processing core. The second layer is the excitation driver 4, with the two DAC outputs connected to Class AB power amplifiers 1 and 2, respectively, driving excitation coil group 1 (A+C area) and excitation coil group 2 (B+D area). The third layer is the object under test (ferromagnetic workpiece), where the excitation coils output excitation signals to the surface of the ferromagnetic workpiece. The magnetic sensor array 5 uses a 3×3 dual-axis (x, z) sensor array to acquire signals fed back from the workpiece. In the signal conditioning module, the sensor signals are processed sequentially through a bias circuit and an instrumentation amplifier before being input to the built-in multi-channel ADC of the microcontroller system 6, forming a closed-loop process.

[0095] This embodiment corresponds to the specific implementation of the hardware composition and mechanical housing structure of the above-mentioned measurement system.

[0096] like Figure 4As shown, the four-region coils are arranged on the PCB board, forming two sets of orthogonal coil groups in pairs. Excitation signal generation: The timer within the microcontroller system triggers DMA to read data from a 256-point sine wave digital array stored in Flash. DAC1 output. DAC2 output ,in, kHz, The amplitude is adjustable (typical value 1.0V).

[0097] like Figures 5-6 As shown, the first coil 7 and the second coil 8 are connected in series to form the first coil group, which is driven by the DAC1 channel through the excitation driver 4; the third coil 9 and the fourth coil 10 are connected in series to form the second coil group, which is driven by the DAC2 channel through the excitation driver 4.

[0098] The driver 4 is a Class AB power amplifier that amplifies the two signals with fixed gain. (e.g., 10 times), output drive coil: , .

[0099] The magnetic sensor array 5 employs a 3×3 array arrangement of TMR / GMR biaxial magnetic sensors. Each sensor detects the x-direction magnetic field component parallel to the workpiece surface and along the direction of the main excitation current, and the z-direction magnetic field component perpendicular to the workpiece surface. The sensors output differential voltage signals. ,in, The sensor sensitivity is given by V / T. Magnetic field data from the magnetic sensors in each x-direction are acquired in real time, and the amplitude and phase of the magnetic field at each point are obtained through digital quadrature demodulation. The average amplitude is then calculated. and phase average Calculate the normalized amplitude deviation for each channel. and phase deviation When the maximum deviation exceeds the set threshold, a proportional control strategy is used to adjust the output amplitude of the corresponding DAC channel. Or the relative phase shift between the two signals Repeat the measurement until the magnetic field uniformity meets the requirements.

[0100] Each sensor output is connected to an instrumentation amplifier for differential amplification and DC biasing, before being output to the ADC. ,in, For instrumentation amplifier gain, It is a DC bias voltage, adapted to the ADC input range of the single-chip microcomputer system 6 (e.g., 0~3.3V).

[0101] The microcontroller system 6 has a built-in ADC that samples the signals of each channel synchronously or in a polling manner at a sampling rate of 100kSPS. The digital signal processing module performs quadrature demodulation, amplitude and phase extraction, closed-loop adjustment and vibration inversion algorithm, and finally outputs vibration data such as vibration displacement, velocity, acceleration, frequency and mode shape.

[0102] like Figure 3 As shown, the entire system is encapsulated within an outer housing 1, with the bottom of the outer housing 1 in close contact with the surface of the ferromagnetic workpiece being tested. The excitation coil assembly is placed at the bottom layer inside the housing and in close contact with the surface of the ferromagnetic workpiece being tested. The magnetic sensor array 5 is placed on top of the excitation coil assembly, the signal conditioning module is placed on top of the magnetic sensor array 5, and the microcontroller system 6 is placed on the top layer. A shielding layer 2 and a heat insulation layer 3 are provided between each layer to suppress the direct coupling of the excitation field to the magnetic sensor array and reduce temperature drift.

[0103] System installation and operation: The bottom surface of the system's outer casing 1 is tightly attached to the surface of the ferromagnetic workpiece being tested. After the system is started, the following process is automatically executed: Establish a rotating magnetic field and complete closed-loop regulation; Entering normal working state, real-time acquisition of ACFM signals from each channel; Complete digital demodulation, strain inversion, and displacement reconstruction; It outputs vibration data such as displacement, velocity, acceleration, frequency, and mode shape.

[0104] Example 8: This embodiment is used to verify the measurement performance of the method and system of the present invention relative to existing technologies (laser Doppler vibration meter and resistance strain gauge).

[0105] 1. Experimental setup The test objects were two rectangular plates of different materials (both measuring 600mm × 400mm × 6mm): Specimen A was a Q345R carbon steel plate; Specimen B was a 304 stainless steel composite plate (Q345R base layer, 304 cladding, total thickness 6mm, cladding thickness 1mm, cladding weakly magnetic, but the ferromagnetism of the base layer was measurable). An electromagnetic vibrator applied sinusoidal excitation to the center of the plate, with frequencies set to 80Hz, 120Hz, 200Hz, and 300Hz, respectively, and the vibration amplitude adjusted to a maximum displacement of approximately 5~10μm on the plate surface.

[0106] The comparison instruments include: Laser Doppler vibration meter (LDV, Polytec OFV-5000) measures the displacement of the center point on the surface of a plate. The resistance strain gauge (BF350-3AA) is attached to the corresponding position at the center of the probe to output the reference strain. 2. Comparison of relative errors in displacement amplitude Using the LDV measurement value as a reference, the displacement amplitude of the central measuring point of this invention was compared, and the results are shown in Table 6 below: Table 6 Comparison of Relative Errors in Displacement Amplitude

[0107] 3. Frequency identification error FFT was performed on the displacement time series measured by the present invention and LDV respectively to extract the main frequency. The frequencies identified by the two were completely consistent with each other, with an error of 0Hz.

[0108] 4. Mode shape correlation coefficient For the four frequencies mentioned above, the correlation coefficients were calculated using the mode shapes reconstructed by the 3×3 array of this invention and the mode shapes obtained by LDV scanning measurements (scanning 9 measurement points). The results are shown in Table 7 below: Table 7 Comparison of Mode Correlation Coefficients

[0109] 5. Linearity and Repeatability Linearity: At a fixed frequency of 120Hz, by changing the vibration amplitude (corresponding to strains of 0.5~50με), the output strain at the center measuring point of this invention is linearly fitted to the reference strain of the resistance strain gauge, and the correlation coefficient is determined. The maximum nonlinear deviation is 2.8%.

[0110] Repeatability: Under the same working conditions (Q345R, 120Hz, displacement amplitude 6.8μm), the measurement was repeated 5 times, the standard deviation of displacement amplitude was 0.12μm, and the coefficient of variation was 1.8%, which meets the engineering repeatability requirements.

[0111] 6. Conclusion Comparative experiments show that, within the frequency range of 80-300Hz, the relative error of displacement measurement for both Q345R carbon steel and 304 composite plate is within ±7%, with zero frequency identification error, a mode shape correlation coefficient greater than 0.88, and linearity and repeatability meeting the requirements for industrial vibration monitoring. Compared with resistance strain gauges, this invention can achieve arrayed measurement without the need for bonding; compared with laser vibrometers, this invention has lower cost and stronger environmental adaptability, verifying the inventiveness and practicality of the solution.

[0112] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for measuring the mechanical vibration of ferromagnetic workpieces based on ACFM, characterized in that, Includes the following steps: S1. Establish a rotating alternating electromagnetic field: Apply a rotating alternating electromagnetic field excitation to the surface of the ferromagnetic workpiece; S2. Closed-loop adjustment of excitation magnetic field: Collect magnetic field data detected by each magnetic field sensor, determine the spatial uniformity of the rotating alternating electromagnetic field, and adjust the excitation parameters when a deviation is detected until the magnetic field uniformity meets the requirements. S3, ACFM signal acquisition: The magnetic field sensor rigidly mounted on the workpiece surface vibrates with the workpiece. The alternating stress / strain generated by the workpiece vibration causes a change in local permeability through the inverse magnetostrictive effect. The change in local permeability modulates the eddy current distribution induced by the rotating alternating electromagnetic field on the workpiece surface, causing the carrier signal detected by each magnetic field sensor to change. After signal conditioning, the discrete signal sequence is acquired by the data acquisition module. S4. Digital demodulation: Perform digital quadrature demodulation on the discrete signals of each channel to extract the amplitude and phase information of each channel signal; S5. Alternating strain inversion: The extracted amplitude information is processed to remove DC and drift to obtain the alternating component caused by vibration. The alternating strain at each measuring point is obtained by inversion using the pre-calibrated comprehensive sensitivity coefficient. S6. Vibration Data Output: Based on the spatial position of the sensor array and known structural parameters, the discrete strain data is spatially reconstructed using the relationship between strain and displacement in bending vibration. The vibration displacement of each measuring point is obtained by solving the problem, and the vibration velocity, acceleration, frequency and mode shape are further calculated to output the vibration data.

2. The method for measuring mechanical vibration of ferromagnetic workpieces based on ACFM according to claim 1, characterized in that, In step S1, the microcontroller system (6) controls the built-in DAC to generate two sinusoidal excitation signals DAC1 and DAC2 with adjustable amplitude and phase difference. After constant amplification by the excitation drive (4), they drive two sets of orthogonal coils respectively to induce a rotating alternating electromagnetic field on the surface of the ferromagnetic workpiece.

3. The method for measuring mechanical vibration of ferromagnetic workpieces based on ACFM according to claim 2, characterized in that, Two orthogonal sinusoidal excitation signals are generated in the following way: a complete cycle of sine wave digital array is pre-stored in the Flash of the microcontroller system (6); the timer triggers DMA transfer to sequentially move the sample point data in the array to the data register of DAC1; the data source of DAC2 is the same as that of DAC1, but the starting reading position is offset from the number of sample points corresponding to a quarter cycle in the array to achieve a 90° phase difference between the two sinusoidal signals.

4. The method for measuring mechanical vibration of ferromagnetic workpieces based on ACFM according to claim 2, characterized in that, In step S2, an amplitude normalized variance threshold and a phase deviation threshold are set. When the amplitude normalized variance or phase deviation of each magnetic field sensor exceeds the corresponding threshold, a proportional control strategy is used to adjust the output amplitude of the corresponding DAC channel or the relative phase offset of the two signals until the magnetic field uniformity meets the requirements.

5. The method for measuring mechanical vibration of ferromagnetic workpieces based on ACFM according to claim 2, characterized in that, In step S4, the microcontroller system (6) synchronously or polls the signals of each channel, generates in-phase reference signals and quadrature reference signals with the excitation frequency as the reference frequency, multiplies them by the sampling sequences of each channel respectively, and then filters out the second harmonic component through a digital low-pass filter to obtain the in-phase component. and orthogonal components ,Depend on and Calculate the amplitude of each channel. and phase time .

6. The method for measuring mechanical vibration of ferromagnetic workpieces based on ACFM according to claim 5, characterized in that, In step S5, for ferromagnetic materials, the contramagnetic-striction effect gives an approximately linear relationship between the change in magnetic permeability and strain: ； in, This represents the change in permeability of a ferromagnetic material under alternating stress / strain. The permeability of ferromagnetic materials, This represents the relative change in magnetic permeability. This is the stress sensitivity coefficient. For elastic modulus, For alternating strain; The relative change in ACFM signal amplitude is proportional to the change in permeability: ； in, The alternating amplitude component of the ACFM detection signal caused by vibration. This is the sensitivity factor of the system to changes in magnetic permeability; By combining the approximately linear relationship between permeability change and strain, and the relationship between the relative change in ACFM signal amplitude and permeability change, a direct relationship between the relative change in amplitude and alternating strain is obtained: ； Among them, the comprehensive calibration coefficient , is a dimensionless constant, which is obtained in one calibration through a standard specimen vibration table experiment; Alternating strain is obtained by inversion from alternating magnetic field data: ； in, This is the average value of the static operating point amplitude; high-pass filtering or moving average is used for DC removal and drift removal.

7. The method for measuring mechanical vibration of ferromagnetic workpieces based on ACFM according to claim 6, characterized in that, In step S6, it is assumed that the workpiece wall is a thin plate structure, bending vibration is dominant, and surface strain and lateral displacement are... The relationship is: ； in, For wall thickness, For the Laplace operator; The spatial reconstruction includes the following two methods: Method 1: When the measured area is far from structural discontinuities, the finite difference method is used for internal nodes. The second derivative is approximated by the central difference: ； in, , The arrays are respectively in direction and Spacing in direction, For nodes Vibration displacement at the location, , They are nodes exist Vibration displacement of adjacent nodes in the direction, , They are nodes exist Vibration displacement of adjacent nodes in the direction; node The relationship between strain and displacement is as follows: ； Write in matrix form , For discrete Laplace matrices, magnetic sensor array The displacement vector is formed by the vibration displacements of each node; the displacement is solved by combining the boundary conditions. Furthermore, a symmetrical extension process is employed to treat the sensor array as the central part of a larger grid, and the strain values ​​of virtual nodes are extrapolated to ensure the well-posedness of the equation system. Method 2: When the area under test is close to a structural discontinuity, the modal superposition method is used: the first few vibration modes of the area under test are obtained in advance through finite element modal analysis or experimental hammer test. and their corresponding strain mode shapes ,in, The modal order index is used. For the first First mode at position The mode shape function at that point, For the first Strain mode shapes of the first mode; solve for the modal coordinates by projecting the measured strain vector onto the strain mode shapes. ,in, For the first First mode with time Reconstructing the displacement field from the changing modal coordinates: 。 8. The method for measuring mechanical vibration of ferromagnetic workpieces based on ACFM according to claim 6, characterized in that, In step S3, the influence of static lift-off distance variation on the measurement results is eliminated through amplitude self-normalization: the amplitude at the static operating point of each channel is used as the standard. For alternating components Normalization yields the relative change. ,in, Channel number, For the first The amplitude alternation component of the channel, For the first The average value of the channel's static operating point amplitude.

9. A mechanical vibration measurement system for ferromagnetic workpieces based on ACFM, applied to the mechanical vibration measurement method for ferromagnetic workpieces based on ACFM as described in any one of claims 1-8, characterized in that, include: The microcontroller system (6) includes a timer, waveform array memory, two DAC channels, an MCU processing core, and a built-in multi-channel ADC. It is used to generate two sinusoidal excitation raw signals DAC1 and DAC2 with adjustable amplitude and phase difference, and to receive the sampled signals from each channel for digital demodulation. The excitation driver (4) is electrically connected to the two DAC channels of the single-chip microcomputer system (6), including a first power amplifier and a second power amplifier, which are used to amplify the two sinusoidal excitation signals with constant gain respectively; The excitation coil group, which is electrically connected to the excitation drive (4), includes a first coil group consisting of a first coil (7) and a second coil (8) connected in series, and a second coil group consisting of a third coil (9) and a fourth coil (10) connected in series, which are used to induce a rotating alternating electromagnetic field on the surface of a ferromagnetic workpiece. A magnetic sensor array (5) is rigidly mounted above the excitation coil group and vibrates with the workpiece. It includes multiple magnetic sensors arranged in an array. Each magnetic sensor detects a magnetic field component parallel to the workpiece surface along the direction of the excitation main current and a magnetic field component perpendicular to the workpiece surface. The signal conditioning module is electrically connected to the magnetic sensor array (5) and includes a bias circuit and an instrumentation amplifier. After amplifying and DC biasing the weak differential signal output by the magnetic sensor array (5), it is input into the built-in multi-channel ADC of the microcontroller system (6).

10. The ACFM-based mechanical vibration measurement system for ferromagnetic workpieces according to claim 9, characterized in that, The system also includes an outer housing (1), the excitation coil group is placed inside the housing at the bottom layer and close to the surface of the ferromagnetic workpiece being tested, the magnetic sensor array (5) is placed on the upper layer of the excitation coil group, the signal conditioning module is placed on the upper layer of the magnetic sensor array (5), and the microcontroller system (6) is placed on the top layer; a shielding layer (2) and a heat insulation layer (3) are provided between each layer to suppress the direct coupling of the excitation field to the magnetic sensor array and reduce temperature drift.

Images