A steel crack automatic detection device and method based on magnetic field orientation control

Abstract

This invention discloses an automatic steel crack detection device and method based on magnetic field-oriented control, relating to the fields of engineering non-destructive testing technology and industrial automation intelligent equipment technology. The device includes: a three-phase rotating magnetic field excitation module, used to generate a synthetic magnetic field with constant intensity and uniform rotation based on a magnetic field-oriented control algorithm, and to magnetize the target material; a magnetic powder atomization module, used to uniformly spray atomized magnetic suspension onto the magnetized target material; an image acquisition module, used to acquire images of the target material after magnetization and spraying; and a main control and processing unit, used to automatically identify cracks in the images acquired by the image acquisition module. This invention enables efficient magnetization, fully automatic and accurate identification, and intelligent analysis in magnetic particle testing, meeting high-standard steel health testing requirements.

Description

Technical Field

[0001] This invention relates to the fields of engineering non-destructive testing technology and industrial automation intelligent equipment technology, and in particular to an automatic detection device and method for steel cracks based on magnetic field orientation control. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As key load-bearing components in urban rail transit, underground utility tunnels, and large-scale public facilities, the health of steel structures directly affects the safety and durability of the overall structure. Steel structures within complex and confined spaces, such as bundled pipe curtains and tunnel support systems, are particularly susceptible to surface and near-surface cracks due to long-term exposure to loads, environmental corrosion, and fatigue stress. However, traditional magnetic particle testing methods still have shortcomings in magnetization efficiency, intelligent identification, and system integration. They are prone to missed detections due to incomplete magnetization coverage, and defect identification currently relies on manual visual interpretation of magnetic traces, resulting in low efficiency and difficulty in ensuring accuracy and traceability, failing to fundamentally resolve the contradiction between testing accuracy and efficiency. Furthermore, for large, regular-surface steel structures such as square bundled pipe curtains in tunnels, existing equipment lacks efficient automated scanning mechanisms. This makes it difficult to achieve high-precision positioning and data acquisition over a large area and in a grid-like manner while ensuring stable contact between the probe and the inspection surface, thus limiting testing efficiency and data comparability. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this invention is to provide an automatic steel crack detection device and method based on magnetic field orientation control, which can achieve efficient magnetization, fully automatic and accurate identification and intelligent analysis of magnetic particle detection, so as to meet the high standards of steel health testing requirements.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: The first aspect of this invention provides an automatic steel crack detection device based on magnetic field orientation control, comprising: The three-phase rotating magnetic field excitation module is used to generate a synthetic magnetic field with constant intensity and uniform rotation based on the magnetic field orientation control algorithm, and to magnetize the detection target. It includes a power drive circuit, a ring common iron core and a three-phase excitation yoke. The ring common iron core is set on the top of the three-phase excitation yoke. The three-phase excitation yoke is connected to three-phase AC power through the power drive circuit to generate a rotating magnetic field and magnetize the detection target. The magnetic powder atomization module is used to uniformly spray the atomized magnetic suspension onto the magnetized detection target; The image acquisition module is used to acquire images of the target after magnetization and spraying. The main control and processing unit is used to automatically identify cracks in the images acquired by the image acquisition module.

[0006] Furthermore, it also includes a mechanical module, which provides mechanical support for the automatic testing equipment. This module includes a detection probe unit and a multi-probe integrated automatic scanning module, with the detection probe unit mounted on the multi-probe integrated automatic scanning module.

[0007] Furthermore, the multi-probe integrated automatic scanning module includes an adaptive support leg system and an adaptive telescopic robotic arm. The adaptive support leg system consists of support legs symmetrically arranged on the four sides (left, right, top, and bottom) of the core host in the central control host module. Each support leg is an independently electrically controllable telescopic mechanical structure with a built-in drive motor and displacement feedback sensor. The adaptive telescopic robotic arm includes a base and a robotic arm. The base is fixed to the top of the connecting column of the central control host module, and the robotic arm is connected to the base. The robotic arm adopts a multi-section telescopic arm and joint combination design, which has multiple degrees of freedom. Driven by the core rotary motor, the robotic arm can achieve 360° positioning and posture determination in the working plane.

[0008] Furthermore, each support leg is equipped with an axial pressure adaptive mechanism to automatically adjust the extension of the support leg.

[0009] Furthermore, a through hole is provided in the center of the annular common iron core, and three three-phase excitation yokes are evenly distributed around the bottom circumference of the annular common iron core as support legs, with an excitation coil tightly wound on each three-phase excitation yoke.

[0010] Furthermore, the power drive circuit includes a DC transformer power supply unit, a control unit, a three-channel reference signal generation unit, and a power amplification unit. The DC transformer power supply unit is used to power various devices. The control unit receives excitation parameters, calculates and stores the three-channel sinusoidal reference voltage data table online. The three-channel reference signal generation unit includes a voltage generation circuit and a signal modulation circuit, which are used to generate reference voltage and map the reference voltage, respectively. The power amplification unit is used to adaptively and dynamically adjust the output current of each group of excitation coils.

[0011] Furthermore, the magnetic powder atomization module includes a magnetic powder liquid nozzle, a liquid guiding tank, and a flow control pump valve connected in sequence. The magnetic suspension is sprayed onto the detection target by the magnetic powder liquid nozzle through the liquid guiding tank, and the flow rate is controlled by the flow control pump.

[0012] Furthermore, the image acquisition module includes a camera, which is mounted in the central through hole of the annular iron core via an adjustable bracket, with the lens facing the detection working surface.

[0013] Furthermore, the main control and processing unit is also used to perform real-time hardware-accelerated preprocessing on the acquired image data; and to run a lightweight convolutional neural network crack recognition model to intelligently analyze and judge the preprocessed images, automatically identify cracks and classify them.

[0014] A second aspect of the present invention provides an automatic detection method for steel crack detection equipment based on magnetic field orientation control as described in the first aspect, comprising the following steps: A synthetic magnetic field with constant intensity and uniform rotation is generated based on a magnetic field orientation control algorithm, and the target to be detected is magnetized. The atomized magnetic suspension is evenly sprayed onto the magnetized target. Acquire images of the target after magnetization and spraying are completed; Automatic crack identification is performed on the acquired images.

[0015] The above one or more technical solutions have the following beneficial effects: This invention discloses an automatic steel crack detection device and method based on magnetic field-oriented control, mainly composed of a hardware system and a software system. The hardware system adopts a robotic configuration, with a central control host module as its core. Its panel is equipped with a multi-functional data interface and a wireless communication module, and internally integrates a high-capacity battery pack and a status indicator unit. An adaptive support leg system serves as a stable base, achieving rapid and active leveling of the device on the workpiece surface through independent electrically controlled extension and retraction of each leg. A multi-degree-of-freedom adaptive telescopic robotic arm serves as the actuator, with its end effector equipped with an integrated detection probe comprising a miniature three-phase excitation magnetic yoke, a magnetic powder nozzle, and a high-definition PCB camera, ensuring precise positioning of the detection components. The software system deeply integrates magnetic field-oriented control, robotic arm trajectory planning, real-time image acquisition and preprocessing, and a crack intelligent recognition algorithm based on a convolutional neural network to achieve closed-loop control throughout the entire process. By incorporating the aforementioned complete set of hardware and software systems, high-precision and rapid detection of steel cracks can be achieved, aiming to solve the problems of slow manual inspection, low accuracy, low magnetization efficiency and poor coverage of traditional magnetic particle inspection, low level of intelligence and accuracy in crack identification, and high power consumption and low efficiency of traditional inspection equipment.

[0016] This invention addresses the problem of low magnetic particle coverage in existing magnetic particle testing equipment, achieving significant improvements in magnetization efficiency and performance. In some embodiments, the three-phase excitation structure used in this invention generates an average magnetic flux density of 9.3 × 10⁻⁶. -6 T is approximately 6.3 × 10⁻⁶ for traditional two-phase excitation. -6The magnetic field strength is 1.5 times that of T. A stronger excitation magnetic field can induce a larger current in the steel being tested, thereby forming a stronger and more comprehensive induced magnetic field. This makes the leakage magnetic field signal generated by microcracks more obvious, significantly improving the defect detection capability and signal-to-noise ratio. Combined with the magnetic field orientation control technology adopted in this scheme, precise control of the phase amplitude of the three-phase current is achieved, further optimizing the uniformity and stability of the magnetic field. While ensuring magnetization coverage, this effectively improves the overall magnetization efficiency and defect detection sensitivity of the system.

[0017] This invention effectively optimizes system power consumption and achieves high energy efficiency. It dynamically adjusts the excitation process through a magnetic field-oriented control algorithm, avoiding over-excitation of the magnetic field and rationally controlling power output while meeting detection requirements. Compared to traditional magnetizing equipment that continuously operates at maximum power, this invention achieves on-demand excitation, improving magnetization intensity while optimizing energy efficiency. This benefits long-term stable operation and battery life in mobile scenarios.

[0018] The intelligent identification and automation levels of the device in this invention are significantly improved. The device integrates magnetic particle visualization enhancement and high-resolution image acquisition technology, combined with auxiliary lighting and ultraviolet light sources, to ensure the capture of high-contrast magnetic trace images under various lighting conditions. The device integrates an embedded processing system based on the ZYNQ platform, accelerating image preprocessing on the FPGA and deploying a lightweight CNN convolutional neural network model on the ARM, enabling real-time, high-precision identification and classification of crack defects, completely replacing the traditional method relying on manual visual interpretation.

[0019] This invention significantly improves the device's adaptability, movement flexibility, and deployment efficiency. It innovatively adopts an integrated robot configuration consisting of a central control host module, an adaptive support leg system, and an adaptive telescopic robotic arm. This configuration allows the device to quickly establish a stable working reference on the workpiece surface through the active extension and retraction of the support legs, without relying on external tracks or precise pre-positioning. Then, the multi-degree-of-freedom dexterity of the robotic arm drives the integrated probe to any detection point, adapting to changes in the curvature of the workpiece surface and maintaining the optimal detection posture at all times. This highly integrated design greatly simplifies pre-detection preparation, eliminates reliance on dedicated scanning tracks, and enables rapid deployment on steel structures of various sizes and shapes (from small pipes to large square pipe curtains). It achieves a leap from "device adapting to the environment" to "device actively adapting to the environment," significantly improving the mobility, adaptability, and overall efficiency of the detection task.

[0020] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the integrated detection probe in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the bottom of the integrated detection probe in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the rotating magnetic field generated by the six continuous-time probes in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the power amplification unit in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of the signal modulation circuit in Embodiment 1 of the present invention; Figure 6 This is a schematic diagram of the automatic detection device in Embodiment 1 of the present invention; Figure 7 This is a magnified front view of a portion of the automatic detection device in Embodiment 1 of the present invention; The components include: 1. Probe caster wheel; 2. Three-phase excitation yoke; 3. Coil wire inlet channel; 4. Excitation coil; 5. Ring common iron core; 6. Wire shorting fixing head; 7. Liquid collection tank; 8. Ring component bracket; 9. Main control and processing unit; 10. Image acquisition module; 11. Main controller; 12. Wire inlet fixing head; 13. Coil wire outlet channel; 14. Camera; 15. Auxiliary lighting device; 16. Magnetic powder atomization module; 17. Support leg caster wheel; 18. Integrated detection probe; 19. Adaptive telescopic support leg system; 20. Adaptive telescopic robotic arm; 21. Core rotary motor; 22. Central control host module; 23. Data transmission module; 24. Charging module; 25. Equipment heat dissipation module; 26. Main control and processing unit. Detailed Implementation

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof. The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0025] Example 1: Embodiment 1 of the present invention provides an automatic steel crack detection device based on magnetic field orientation control, including a mechanical module, a three-phase rotating magnetic field excitation module, a magnetic powder atomization module, an image acquisition module, and a central control host module.

[0026] The mechanical module provides mechanical support for the automatic testing equipment and includes a detection probe unit and a multi-probe integrated automatic scanning module. The detection probe unit is mounted on the multi-probe integrated automatic scanning module.

[0027] In one specific implementation, such as Figure 1 and Figure 2 As shown, the detection probe unit includes an integrated detection probe 18 and probe casters 1. The integrated detection probe 18 compactly encapsulates a miniaturized three-phase rotating magnetic field excitation module, a magnetic powder atomization module, and an image acquisition module. The probe casters 1 are mounted at the ends of the three-phase excitation yoke 2, and can be moved flexibly within the detection area of ​​the workpiece by means of an adaptive telescopic robotic arm. A housing can be provided outside the detection probe unit, with a plastic encapsulated housing covering the integrated probe and fixed to a ring-shaped component bracket.

[0028] The multi-probe integrated automatic scanning module includes an adaptive support leg system 19 and an adaptive telescopic robotic arm 20, such as Figure 6 and Figure 7As shown, the adaptive support leg system 19 consists of support legs made of aluminum alloy, steel pipe, or rigid plastic pipe, symmetrically arranged on the four sides (left, right, top, and bottom) of the core host in the central control host module. Two support legs are installed on each side, perpendicular to the side, located at both ends of each side, and fixed to the core host with screws. The ends of the support legs are equipped with casters 17, allowing the entire device to move freely within the inspected workpiece. The drive force can be provided manually.

[0029] To accommodate the flatness or slight deformation of the steel structure surface, each support leg is equipped with an axial pressure adaptive mechanism (such as a spring buffer module) to automatically adjust the extension of the support leg. This ensures that the support leg maintains a constant optimal contact pressure with the steel structure surface during the scanning process, guaranteeing the consistency of the detection conditions and enhancing the equipment's adaptability to detection tasks with different bundled tube curtain diameters.

[0030] Each support leg is an independently electrically controlled telescopic mechanical structure with a built-in drive motor and displacement feedback sensor. The system can automatically adjust the extension length of each leg according to the surface condition of the workpiece being inspected (such as a square bundled tube curtain), so that the central control host module is stably supported inside the workpiece being inspected, and provides a stable reference platform for subsequent inspection.

[0031] The adaptive telescopic robotic arm 20 includes a base and a robotic arm. The base is fixed to the top of the connecting column of the central control host module, and the robotic arm is connected to the base. The robotic arm adopts a multi-section telescopic arm and joint combination design, possessing multiple degrees of freedom. Driven by the core rotary motor, the robotic arm achieves 360° positioning and posture within the working plane. A detection probe unit is mounted at the end of the robotic arm, which is responsible for precisely moving the probe to the preset detection point.

[0032] The three-phase rotating magnetic field excitation module is used to generate a synthetic magnetic field with constant intensity and uniform rotation based on the magnetic field orientation control algorithm, and to magnetize the detection target.

[0033] In one specific implementation, such as Figure 1 and Figure 2As shown, the three-phase rotating magnetic field excitation module includes a power drive circuit, a ring-shaped common iron core, and a three-phase excitation yoke. The ring-shaped common iron core is located on top of the three-phase excitation yoke. The three-phase excitation yoke is connected to three-phase AC power through the power drive circuit to generate a rotating magnetic field, which magnetizes the detection target. The ring-shaped common iron core 5 has a through hole in its center to leave space for the optical system. Three three-phase excitation yokes are evenly distributed around the bottom circumference of the ring-shaped common iron core as support legs. Each three-phase excitation yoke has an excitation coil 4 tightly wound on it. It is connected to three-phase AC power through the power drive circuit to generate a composite magnetic field with constant intensity and uniform rotation, achieving one-time all-round efficient magnetization of the workpiece, solving the problems of traditional single-phase magnetic yokes requiring multiple operations and having poor coverage.

[0034] The three-phase rotating magnetic field excitation module also includes a ring component bracket and a fixed disk. The ring component bracket is mounted on the fixed disk where three magnetic yokes are in contact, and is used to fix the various components. The fixed disk is configured as three concentric rings, namely an outer ring, a middle ring, and an inner ring from the outside to the inside. The outer ring fixes three excitation magnetic yokes, the middle ring fixes the magnetic powder atomization module, and the inner ring fixes the camera and auxiliary light source device. Among them, the middle ring has six through holes for placing the magnetic powder liquid nozzle. In this embodiment, in order to achieve uniform atomization, a 120° triangularly distributed magnetic powder liquid nozzle is preferably used.

[0035] The coil wire inlet channel 3 and the coil wire outlet channel 13 are located inside the three excitation yokes, respectively, and are used for the entry and exit of the wires. They can also concentrate the coil wires at the wire shorting fixing head 6 and the wire access fixing head 12. The wire shorting fixing head 6 and the wire access fixing head 12 are mounted on the annular component bracket 8 and are used to realize the shorting of the wires and the current access, respectively.

[0036] Specifically, the working principle of the three-phase rotating magnetic field excitation module is as follows: Three-phase symmetrical sinusoidal alternating currents of equal amplitude, same frequency, and phases differing by 120 degrees are passed through the three-phase coils: , , .

[0037] in, , , The currents for phases a, b, and c are respectively. The current amplitude, Angular frequency, For driving frequency, Given time, the rotation period of the rotating magnetic field can be obtained. .

[0038] like Figure 3As shown, in the target plane below the coil center region, neglecting edge effects, the magnetic induction intensity generated by each phase coil can be decomposed into two orthogonal components in the x and y directions along this plane. Due to the symmetrical design of the coil, the magnetic field generated by a single-phase coil on the target plane can be represented as a fixed spatial vector proportional to the phase current. Combining the contributions of all three phases, the total magnetic induction intensity B at a point on the target plane within the coverage area of ​​the three-phase coil can be expressed as... .

[0039] Where B(t) is the total magnetic field, The magnetic field amplitude coefficient is determined by the coil geometry, number of turns, and current amplitude. , , This is a weighted vector representing the spatial direction and relative intensity of the magnetic fields generated by the a, b, and c phase coils at the target point. Through coil layout design, these three spatial vectors are arranged at 120 degrees to each other and have equal lengths within the target plane, satisfying the following condition: .

[0040] The total magnetic field B(t) can be simplified to a vector rotating uniformly in the target plane: .

[0041] in, Let be the amplitude of the rotating magnetic field, and k be a weighting factor, representing the relative strength of the magnetic field generated by each phase coil at the target point. and Let be two orthogonal unit vectors in the target plane. The rotational angular velocity of these vectors is... The direction is determined by the phase sequence of the current.

[0042] The power drive circuit includes a DC transformer power supply unit, a control unit, a power amplifier unit, and a three-channel reference signal generation unit. The DC transformer power supply unit supplies power to all devices, connecting all modules or units in the detection equipment that require power. The control unit is connected to the three-channel reference signal generation unit, receiving excitation parameters (pre-set by the user), calculating and storing three sinusoidal reference voltage data tables online. The three-channel reference signal generation unit is connected to the power amplifier unit, generating corresponding reference voltages according to the control unit's instructions. The three-channel reference signal generation unit includes a voltage generation circuit and a signal modulation circuit, used to generate the reference voltage and map it, respectively. The power amplifier unit adaptively and dynamically adjusts the output current of each excitation coil, amplifying the reference voltage to drive the excitation coils to generate a rotating magnetic field for magnetizing the detection target.

[0043] Specifically, the DC transformer power supply unit centrally designs and manages the voltages used in the system. To meet the actual engineering testing needs, a 12V lithium battery is used for power supply, and an internal DC-DC transformer high-frequency switching boost circuit generates a 30V DC main positive voltage output for powering the power amplifier unit, a ±12V DC auxiliary voltage for powering the analog circuit, and a 5V DC auxiliary power supply voltage for powering the control unit and the main control and processing unit.

[0044] Specifically, the control unit uses an embedded 32-bit microcontroller with an ARM core. It receives excitation parameters such as frequency and amplitude set by the user, calculates three sinusoidal reference voltage data tables online, and stores them in the on-chip RAM. The calculation formula is as follows: , , .

[0045] In the formula, , , The table contains sinusoidal reference voltage data for the coils of phases a, b, and c, respectively. 'i' represents the i-th data point in the voltage data table, and 'size' represents the number of data points in the table. Frequency control is achieved by triggering data transmission using a timer. The formula for calculating the timer trigger frequency is: .

[0046] in, For timer trigger frequency, The clock frequency is the timer peripheral of the microcontroller, and PSC is the prescaler setting for the timer peripheral. The automatic reload value set for the timer peripheral.

[0047] Specifically, the power amplification unit consists of a high-current operational amplifier circuit and a signal pre-stage buffer circuit. To avoid magnetic field control errors caused by inconsistencies in the electrical parameters of the three sets of excitation coils, a current feedback path is designed to ensure that the module's output current value always follows changes in the input voltage value (i.e., the voltage-controlled current function described later), thereby adaptively and stably driving each set of excitation coils. The design principle is as follows: Figure 4 As shown, it consists of a voltage follower, a voltage-controlled current source, and a clamping protection circuit. The voltage follower, composed of a low-noise operational amplifier U2 (OPA445), resistors R5 and R6, buffers and isolates the input signal Vcur_A_in to increase the input impedance and avoid subsequent load effects. The voltage-controlled current source consists of a high-voltage, high-current operational amplifier U3 (OPA549), a positive feedback loop composed of current-limiting resistors R9, R7, R10, and C2, and a high-precision, low-temperature drift resistor network (R11, R12, R13, R14) connected to R8 to control the input signal. The output signal Vcur_A_out is processed to have high current driving capability and follow load changes. The clamping protection circuit consists of two Schottky diodes, D1 and D2. They conduct when the output voltage is higher than VCCi, limiting the upper limit of the output; and conduct when the output voltage is lower than VSSi, limiting the lower limit of the output. This module simultaneously introduces positive and negative feedback loops and a high-precision, low-temperature drift resistor network to form the module's output current feedback path. It achieves voltage-controlled current function in a simple circuit, offering advantages such as simple and reliable circuitry, low failure rate, minimal external interference, and easy debugging.

[0048] Based on the inherent characteristics of operational amplifiers, they exhibit "virtual short" and "virtual open" properties when operating in the linear region. Therefore, the following calculations can be performed. According to the "virtual open" property, we can obtain: , .

[0049] because , Therefore, we can conclude that: , .

[0050] Based on the nature of "virtual shortness", : .

[0051] At the same time, according to Kirchhoff's current law, we can obtain: .

[0052] Combining the above equations, we get: .

[0053] Maximum output voltage of the power amplifier unit = VCC, therefore, output node ;at the same time, and ,therefore This can be ignored. Therefore, the output current of this unit can be obtained. With input reference voltage The relation is: .

[0054] when At that time, it can be obtained .

[0055] By selecting different resistance values, the output current is controlled to maintain a fixed proportional relationship with the input voltage, thereby enabling the drive current of each excitation coil to follow the output despite external interference, and thus achieving accurate control of the magnetic field.

[0056] Specifically, the three-channel reference signal generation unit consists of a voltage generation circuit composed of two DAC8562 chips and a signal modulation circuit composed of three LM358 chips. The voltage generation circuit outputs a reference voltage from 0V to 5V and connects to the control unit via a multiplexed SPI bus. The control unit writes data into the voltage data table, causing the three reference signal generation units to sequentially output specified analog reference voltage values. To ensure consistency in the triggering time of the three reference voltages, two DAC8562 chips are configured as external LDAC pin trigger outputs. The signal modulation circuit maps the 0V-5V analog voltage reference output from the voltage generation circuit to a range of -5V to 5V. The design principle of the signal modulation circuit is as follows: Figure 5 As shown in the diagram, LM358 is a high-gain operational amplifier with internal frequency compensation. R1 is a damping resistor, R2 and R3 form an input voltage resistor divider network, R4 is a feedback resistor, R5 is an output buffer resistor, and C1 is a feedback compensation capacitor. Similarly, based on the inherent characteristics of operational amplifiers—"virtual short" and "virtual open"—the relationship between the output voltage and the input reference voltage of this unit can be calculated as follows: .

[0057] make When the voltage is applied, a reference voltage ranging from -5V to +5V can be output.

[0058] in, This represents the voltage at the inverting input terminal of the entire amplifier circuit. This represents the voltage at the non-inverting input terminal of the entire amplifier circuit. This represents the output voltage of the entire amplifier circuit.

[0059] It should be noted that the circuit structures of the three reference signal generation units in this embodiment are all composed of existing circuit structures, which will not be described in detail here.

[0060] The magnetic powder atomization module 16 is used to uniformly spray the atomized magnetic suspension onto the magnetized detection target.

[0061] In one specific embodiment, the magnetic powder atomization module 16 includes a magnetic powder liquid nozzle, a liquid guiding tank, and a flow control pump valve connected in sequence. The magnetic suspension in the collection tank 7 is sprayed onto the detection target through the magnetic powder liquid nozzle via the liquid guiding tank, and the flow rate is controlled by the flow control pump. The nozzle direction of the magnetic powder liquid nozzle is optimized to ensure that the magnetic suspension can uniformly cover the inspected area. The module's start and stop times are precisely controlled by the main controller, working in conjunction with the magnetization process to achieve the best magnetic powder aggregation and display effect at the defect location.

[0062] Image acquisition module 10 is used to acquire images of the target magnetization and after the spraying is completed; In one specific embodiment, the image acquisition module 10 includes a camera 14, which is a high-definition PCB camera mounted in the central through-hole of the toroidal iron core via an adjustable bracket, with the lens facing the inspection work surface. The image acquisition camera employs a high-density strobe light source and a high-precision electronically controlled lens, and is equipped with an auxiliary lighting device 15 to adapt to different lighting environments, ensuring that high-contrast, clear magnetic trace images are captured after magnetic powder spraying. The auxiliary lighting device 15 consists of an LED supplementary light source and an optional ultraviolet light source, mounted on the inner ring of the fixed disk and fixed by threads, facing the inspected workpiece. This module can still guarantee image quality in complex environments, providing reliable input for subsequent analysis.

[0063] The central control host module 22, serving as the main body and control core of the equipment, includes a core host and connecting columns. The core host consists of two identical cuboids connected by screws via cylindrical connecting columns. Figure 7 As shown. A core rotary motor 21 is also installed inside the connecting column to provide driving force for the adaptive robotic arm. The core host is an integrated enclosure with a front panel on its front surface. The front panel houses a data transmission module 23, which includes multiple USB / Type-C data interfaces, a VGA display interface, a wired network interface, and a wireless communication module for data interaction, external device connection, and remote communication. The core host also integrates a charging module 24 and a device heat dissipation module 25. The charging module 24 includes a high-capacity battery pack to provide power to the entire system. The device heat dissipation module 25 is used to cool the entire core host. The front panel also features system status indicator lights to indicate the device's operating mode, battery level, and communication status in real time. The main control and processing unit 9 is used for automatic crack identification of images acquired by the image acquisition module.

[0064] In one specific implementation, the main control and processing unit 9 is embedded within the core host, which utilizes a ZYNQ series embedded processor. This chip integrates both FPGA and ARM architectures. The FPGA side is responsible for implementing the magnetic field orientation control algorithm and performing real-time hardware-accelerated preprocessing of the image data acquired by the camera. The ARM side runs a lightweight convolutional neural network crack recognition model to intelligently analyze and judge the preprocessed image. The main control and processing unit 9 is also used for real-time hardware-accelerated preprocessing of the acquired image data; running a lightweight convolutional neural network crack recognition model to intelligently analyze and judge the preprocessed image, automatically identifying and classifying cracks.

[0065] After the equipment is powered on, the main controller 11 (ZYNQ platform) embedded in the central control host module starts up. The system main control program running on the ARM side first performs initialization operations, loads the pre-trained CNN crack recognition model into memory, and completes handshake communication and self-test with the magnetic field orientation controller, robotic arm motion controller, image acquisition controller, magnetic powder liquid pump valve controller, and adaptive support leg drive controller through the internal communication bus. After all modules report normal status, the equipment enters standby mode. Upon receiving the start detection command, the four-way adaptive support leg system acts first. Each support leg extends and retracts independently according to the control strategy, so that the central control host module is stably supported on the surface of the workpiece being inspected (such as a square bundled tube curtain), establishing a stable operating reference for the robotic arm. Subsequently, the multi-degree-of-freedom adaptive telescopic robotic arm starts working, moving the integrated detection probe at its end to the first point of the initial path dot matrix generated by the main controller according to the detection area. The probe automatically fine-tunes its posture through sensor feedback to ensure that the probe end face maintains the optimal normal contact relationship with the surface being inspected.

[0066] After confirming the probe is in the optimal detection posture, the main controller immediately triggers the magnetic field orientation control algorithm. Through the power drive circuit, a three-phase current with a 120-degree phase difference is injected into the miniature three-phase excitation coil inside the probe, generating a composite magnetic field with constant intensity and uniform rotation beneath the workpiece surface. This achieves efficient, all-around magnetization of the detection point in a single operation. Almost simultaneously, the main controller instructs the magnetic powder pump valve to start, uniformly and atomizingly spraying the magnetic suspension through a nozzle on the probe onto the magnetized area. The leakage magnetic field at the defect immediately attracts magnetic powder, forming a clear magnetic trace. Following this, the high-definition PCB camera integrated on the probe, with the assistance of the auxiliary lighting module, quickly captures the magnetic trace image at the current location. The acquired raw image data is sent at high speed to the FPGA of the ZYNQ chip, where real-time preprocessing, including grayscale conversion, noise reduction, and contrast enhancement, is performed using hardware parallel computing capabilities to improve image quality.

[0067] During the magnetization process, the main controller synchronously instructs the magnetic powder liquid pump valve controller to start, precisely opening the flow control pump valve to uniformly and atomize the magnetic suspension in the collection tank and spray it onto the workpiece surface through the magnetic powder liquid nozzle. Under the influence of the leakage magnetic field at the defect, the magnetic powder is rapidly attracted and aggregated, forming a magnetic mark visible to the naked eye.

[0068] After magnetization and spraying are completed, the main controller triggers the image acquisition controller, which controls the PCB camera to capture high-definition images of the workpiece surface with the assistance of the lighting module. The acquired raw image data is first sent at high speed to the FPGA of the ZYNQ chip, where a series of image preprocessing operations are performed in parallel using hardware logic, including grayscale conversion, noise reduction, and contrast enhancement, which greatly improves image quality and reduces the computational burden for subsequent intelligent recognition.

[0069] The preprocessed image data is transmitted to the ARM processor of the ZYNQ chip. The convolutional neural network (CNN) inference engine running on the ARM performs real-time forward propagation calculations on the image, automatically identifying and locating magnetic indentation regions, determining the presence of cracks or defects, and performing preliminary classification of defect types (such as cracks, porosity, and inclusions) based on the model's capabilities. The identification results are accompanied by confidence scores, ensuring the objectivity and accuracy of the interpretation, replacing the traditional method that relies on manual visual inspection.

[0070] After intelligent identification is completed, the system automatically generates an inspection report. The report may include the presence, location, type, and confidence level of defects. The results can be wirelessly transmitted to a remote host computer or cloud platform via a communication interface (4G / 5G or Wi-Fi, etc.) for storage, visualization, and further big data analysis. At the same time, control signals can also be output through the IO interface to directly trigger subsequent execution mechanisms such as sorting and alarms, forming a complete automated inspection closed loop.

[0071] The implementation method of this invention includes three major stages: system initialization and self-testing, fully automated detection process execution, and intelligent recognition and decision-making. It achieves full automation from workpiece magnetization, magnetic particle display, image acquisition to AI intelligent judgment, significantly improving the efficiency, accuracy, and intelligence level of steel crack detection. Through deep collaboration between hardware and software systems, high-precision and high-efficiency automated detection of steel cracks is achieved.

[0072] Example 2: Embodiment 2 of the present invention provides an automatic detection method for steel crack automatic detection equipment based on magnetic field orientation control as described in Embodiment 1, comprising the following steps: A synthetic magnetic field with constant intensity and uniform rotation is generated based on a magnetic field orientation control algorithm, and the target to be detected is magnetized. The atomized magnetic suspension is evenly sprayed onto the magnetized target. Acquire images of the target after magnetization and spraying are completed; Automatic crack identification is performed on the acquired images.

[0073] The steps and methods involved in the above embodiment two correspond to those in embodiment one. For specific implementation details, please refer to the relevant description section of embodiment one.

[0074] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0075] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0076] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An automatic steel crack detection device based on magnetic field orientation control, characterized in that, include: The three-phase rotating magnetic field excitation module is used to generate a synthetic magnetic field with constant intensity and uniform rotation based on the magnetic field orientation control algorithm, and to magnetize the detection target. It includes a power drive circuit, a ring common iron core and a three-phase excitation yoke. The ring common iron core is set on the top of the three-phase excitation yoke. The three-phase excitation yoke is connected to three-phase AC power through the power drive circuit to generate a rotating magnetic field and magnetize the detection target. The magnetic powder atomization module is used to uniformly spray the atomized magnetic suspension onto the magnetized detection target; The image acquisition module is used to acquire images of the target after magnetization and spraying. The main control and processing unit is used to automatically identify cracks in the images acquired by the image acquisition module.

2. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 1, characterized in that, It also includes a mechanical module, which provides mechanical support for the automatic testing equipment. This module includes a detection probe unit and a multi-probe integrated automatic scanning module, with the detection probe unit mounted on the multi-probe integrated automatic scanning module.

3. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 2, characterized in that, The multi-probe integrated automatic scanning module includes an adaptive support leg system and an adaptive telescopic robotic arm. The adaptive support leg system consists of support legs symmetrically arranged on the four sides (left, right, top, and bottom) of the core host in the central control host module. Each support leg is an independently electrically controllable telescopic mechanical structure with a built-in drive motor and displacement feedback sensor. The adaptive telescopic robotic arm includes a base and a robotic arm. The base is fixed to the top of the connecting column of the central control host module, and the robotic arm is connected to the base. The robotic arm adopts a multi-section telescopic arm and joint combination design, which has multiple degrees of freedom. Driven by the core rotary motor, the robotic arm can achieve 360° positioning and posture determination in the working plane.

4. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 3, characterized in that, Each support leg is equipped with an axial pressure adaptive mechanism to automatically adjust the extension of the support leg.

5. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 1, characterized in that, The toroidal common core has a through hole in the center, and three three-phase excitation yokes are evenly distributed around the bottom circumference of the toroidal common core as support legs, with excitation coils tightly wound on each three-phase excitation yoke.

6. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 1, characterized in that, The power drive circuit includes a DC transformer power supply unit, a control unit, a three-channel reference signal generation unit, and a power amplification unit. The DC transformer power supply unit is used to power various devices. The control unit receives excitation parameters, calculates and stores the three-channel sinusoidal reference voltage data table online. The three-channel reference signal generation unit includes a voltage generation circuit and a signal modulation circuit, which are used to generate reference voltage and map the reference voltage, respectively. The power amplification unit is used to adaptively and dynamically adjust the output current of each group of excitation coils.

7. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 1, characterized in that, The magnetic powder atomization module includes a magnetic powder liquid nozzle, a liquid guide tank, and a flow control pump valve connected in sequence. The magnetic suspension is sprayed onto the detection target by the magnetic powder liquid nozzle through the liquid guide tank, and the flow rate is controlled by the flow control pump.

8. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 1, characterized in that, The image acquisition module includes a camera, which is mounted in the central through hole of the ring-shaped iron core via an adjustable bracket, with the lens facing the detection working surface.

9. The automatic steel crack detection equipment based on magnetic field orientation control as described in claim 1, characterized in that, The main control and processing unit is also used to perform real-time hardware-accelerated preprocessing of the acquired image data; it runs a lightweight convolutional neural network crack recognition model to intelligently analyze and judge the preprocessed images, and automatically identify and classify cracks.

10. An automatic detection method for steel crack automatic detection equipment based on magnetic field orientation control as described in any one of claims 1-9, characterized in that, Includes the following steps: A synthetic magnetic field with constant intensity and uniform rotation is generated based on a magnetic field orientation control algorithm, and the target to be detected is magnetized. The atomized magnetic suspension is evenly sprayed onto the magnetized target. Acquire images of the target after magnetization and spraying are completed; Automatic crack identification is performed on the acquired images.

Images