High-speed camera monitoring method and system for electromagnetic dynamic dual-axis synchronous loading

Abstract

This invention discloses a high-speed camera monitoring method and system for electromagnetic dynamic biaxial synchronous loading, belonging to the field of aerospace component performance testing technology. The turbine engine rotor component to be monitored is mounted on an electromagnetic dynamic biaxial loading test bench. Radial and axial electromagnetic loaders are driven to simulate alternating load conditions during actual engine operation. A fracture excitation device is activated to cause instantaneous fracture at a predetermined location, releasing fragments and acquiring a sequence of fragment flight images. The two-dimensional images are unified into three-dimensional space, and the three-dimensional flight trajectory coordinate sequence of the fragments and the deformation boundary line of the casing inner wall are extracted. The incident velocity vector of the fragment impacting the casing inner wall and the exit velocity vector after penetration are calculated to determine whether the fragment has penetrated the casing. This invention can realistically reproduce the fracture and dispersion process of high-speed, high-energy fragments under biaxial dynamic loading, improving the accuracy of three-dimensional trajectory measurement and the reliability of penetration determination, providing accurate experimental basis for the containment design and airworthiness verification of aero-engine casings.

Description

Technical Field

[0001] This invention discloses a high-speed camera monitoring method and system for electromagnetic dynamic dual-axis synchronous loading, belonging to the field of aviation component performance testing technology. Background Technology

[0002] During operation, components of an aircraft turbine engine, such as rotor blades and discs, may be impacted by foreign objects like birds, hail, or debris, or develop fatigue cracks due to long-term service, eventually fracturing under immense centrifugal force. If the high-speed, high-energy debris generated after fracture is effectively contained within the engine casing, the hazard can be limited to the engine's interior. However, if the casing fails to contain the debris, it will penetrate and fly out, potentially striking critical structural components such as the aircraft fuselage, fuel tanks, hydraulic lines, electrical wiring, or flight control systems. This could lead to serious malfunctions such as cabin depressurization, system failures, and equipment malfunctions, or even catastrophic aviation accidents. Therefore, accurately assessing the engine casing's ability to contain fractured debris during the design phase, and precisely monitoring the debris's trajectory, impact location, velocity changes, and penetration behavior within the casing, are crucial for ensuring flight safety.

[0003] Currently, simulation test methods for the movement of fragments after engine rotor component fracture typically employ a mechanical or pneumatic single-shaft loading device to apply static or quasi-static preload to the rotor component, then cause the component to fracture through mechanical impact or explosion, and finally use a high-speed camera system to capture the fragment scattering process.

[0004] However, such methods have significant shortcomings: Firstly, traditional loading devices struggle to simultaneously simulate the dynamic coupling effect of radial centrifugal force and axial aerodynamic thrust experienced by the rotor under real engine operating conditions, and can only apply loads in a single direction or at a constant rate, resulting in significant deviations between the fracture initiation location, initial debris velocity, and dispersion direction and the actual situation. Secondly, existing high-speed camera monitoring methods are mostly single-view or asynchronous acquisition, unable to accurately acquire the complete flight trajectory of debris in three-dimensional space and the deformation boundary of the casing inner wall at the moment of impact, making it even more difficult to accurately distinguish whether debris bounces back into the casing or penetrates the casing and flies outward. Furthermore, insufficient synchronization accuracy between the fracture trigger moment and camera acquisition often leads to missing keyframes or motion blur, making it difficult to obtain reliable incident and ejection velocity vectors for subsequent analysis. Summary of the Invention

[0005] To overcome the aforementioned shortcomings, this invention proposes a high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading. This method employs radial and axial electromagnetic loaders driven by independent timing controllers to apply frequency and phase-adjustable dual-axis dynamic electromagnetic forces to the rotor components, simulating the alternating load conditions during real engine operation. Multiple high-speed camera groups are synchronously triggered via equal-length cables, and combined with narrowband filtering and pulse stroboscopic illumination technology, high-definition images are continuously acquired starting at the moment of fracture. Furthermore, a spatial coordinate transformation matrix is ​​used to unify the multi-view two-dimensional images into a three-dimensional coordinate system, accurately calculating the debris's flight trajectory, incident velocity vector, and post-impact ejection velocity vector, thereby reliably determining whether the debris has penetrated the casing and recording the remaining velocity direction. This method can provide real, accurate, and repeatable test data for the containment design and airworthiness verification of aero-engine casings.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] A high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading includes:

[0008] S1: The turbine engine rotor component to be monitored is installed on the rotor support structure of the electromagnetic dynamic biaxial loading test bench. Radial electromagnetic loaders are arranged in the radial direction and axial electromagnetic loaders are arranged in the axial direction of the test bench. Multiple high-speed camera groups with optical spindles pointing to the preset debris scattering space area are arranged around the outer casing.

[0009] S2: By sending loading commands to the radial electromagnetic loader and the axial electromagnetic loader respectively through an independent timing controller, the two electromagnetic loaders synchronously apply dynamically changing electromagnetic forces to generate a dual-axis synchronous preload on the rotor component that simulates the operating conditions of an engine.

[0010] S3: Activate the fracture excitation device installed at the predetermined fracture site of the rotor component, so that the predetermined fracture site will fracture instantaneously under the combined action of biaxial preload, releasing multiple fragments.

[0011] S4: The timing controller sends synchronous trigger pulses to all high-speed camera groups to acquire images from the moment the fragment breaks until the fragment hits the inner wall of the casing and stops moving, recording the instantaneous image of the fragment in space and the deformation stripe image of the inner wall surface of the casing.

[0012] S5: Extract the geometric centroid coordinates of the fragments and the deformation boundary line of the inner wall of the casing frame by frame from the image sequence, and use the pre-calibrated spatial coordinate transformation matrix to obtain the flight trajectory coordinate sequence of the fragments in three-dimensional space and the boundary of the impact indentation area of ​​the inner wall of the casing.

[0013] S6: Calculate the incident velocity vector and the exit velocity vector after penetration at the impact point of the fragment on the inner wall of the casing based on the flight trajectory coordinate sequence. Determine whether the fragment has penetrated the casing based on whether the exit velocity vector is zero, and record the direction of the remaining velocity vector when penetration occurs.

[0014] Furthermore, the radial electromagnetic loader described in S1 includes a plurality of radial electromagnets evenly distributed along the circumference of the rotor component, the core axis of each radial electromagnet coincides with the radial direction of the rotor component, and a non-contact air gap is maintained between the core end face of each radial electromagnet and the outer edge surface of the rotor component.

[0015] The axial electromagnetic loader includes at least two coaxially arranged annular electromagnets, which are respectively disposed on the front end face and the rear end face of the rotor component. The inner diameter of each annular electromagnet is larger than the diameter of the rotor shaft of the rotor component. The end face of the annular electromagnet and the corresponding end face of the rotor component also maintain a non-contact air gap.

[0016] The inner wall surface of the outer casing is coated with a diffuse reflection coating, which is formed by spraying a mixture of magnesium oxide powder and epoxy resin in a certain proportion, and is used to enhance the texture contrast of the inner wall of the casing in high-speed camera images.

[0017] Furthermore, the dynamically changing electromagnetic force described in S2 is a time-varying force in the form of a sine wave. The electromagnetic forces applied by the radial electromagnetic loader and the axial electromagnetic loader have the same oscillation frequency, and the phase difference between the electromagnetic forces in the two directions is preset according to the force phase relationship of the rotor components in actual engine operation.

[0018] The timing controller contains two independent waveform generation channels. Each channel outputs a programmable voltage waveform to the drive coil of the corresponding electromagnetic loader. The two channels are locked in phase by a clock synchronization signal, so that the radial electromagnetic force and the axial electromagnetic force can change in tandem according to a preset phase difference. During the application of the dual-axis synchronous preload, the magnitude of the electromagnetic force gradually increases from zero to the target amplitude. The rate of increase is determined according to the material strength and fracture toughness of the rotor components to avoid unexpected premature fracture before the fracture excitation device is activated.

[0019] Furthermore, S4 specifically includes the following sub-steps:

[0020] S41: Generates rising edge synchronous trigger pulses inside the timing controller and transmits them in parallel to the external trigger interface of each high-speed camera group via coaxial cable;

[0021] S42: Each high-speed camera group is equipped with a high-precision crystal oscillator. When the external trigger interface receives the rising edge of the synchronous trigger pulse, the crystal oscillator starts counting and starts the first frame exposure of the camera group after reaching the preset delay count. The delay count is independently calibrated according to the optical path distance between each camera group and the fracture excitation position.

[0022] S43: Each high-speed camera group continuously performs exposure and readout operations for subsequent frames according to a pre-set fixed frame rate. The interval between the exposure start times of two adjacent frames is obtained by frequency division of the crystal oscillator. The fixed frame rate is determined based on the expected maximum speed of the debris flying in the internal space of the housing and the maximum allowable motion blur displacement of the debris in each frame image.

[0023] S44: A narrow-band filter is installed in front of the photosensitive surface of the image sensor of each high-speed camera group. The center wavelength is matched with the emission wavelength of the stroboscopic illumination source arranged around the test bench. During the exposure of each frame image, the timing controller sends a pulse drive current synchronized with the exposure window of the camera group to the stroboscopic illumination source. The edge contour of the fragment and the deformation characteristics of the inner wall surface of the casing are recorded.

[0024] S45: Each high-speed camera group continuously acquires images until it detects that the change in the position of a fragment in multiple consecutive frames is less than the spatial distance corresponding to one pixel. It then automatically stops acquiring images and stores the acquired complete image sequence in the non-volatile memory inside the high-speed camera group.

[0025] Furthermore, the pre-calibrated spatial coordinate transformation matrix described in S5 is obtained in the following way:

[0026] A calibration target is placed in the debris scattering area, and multiple high-speed camera groups simultaneously capture images of the calibration target to extract the two-dimensional pixel coordinates of the spherical marker points.

[0027] For each camera group, establish the projection relationship between the two-dimensional pixel coordinates extracted from the image of the group and the three-dimensional coordinates of the corresponding marker points on the calibration target, and solve for the projection matrix corresponding to the group.

[0028] By combining the projection matrices of all camera groups into a set of three-dimensional spatial coordinate transformation equations, the transformation matrix from the two-dimensional pixel coordinates of any camera group to the unified three-dimensional spatial coordinates can be obtained.

[0029] Furthermore, the incident velocity vector described in S6 is calculated in the following manner:

[0030] The three-dimensional coordinates of the frame before the fragment contacts the inner wall of the casing and the three-dimensional coordinates of the frame before the contact are selected from the fragment flight trajectory coordinate sequence. The difference vector is divided by the time interval between the two frames to obtain the incident velocity vector.

[0031] The ejection velocity vector is calculated in the following way:

[0032] Select the three-dimensional coordinates of the first frame after the fragment contacts the inner wall of the casing and the three-dimensional coordinates of the second frame after contact from the fragment flight trajectory coordinate sequence. Divide the difference vector by the time interval between the two frames to obtain the ejection velocity vector. If the three-dimensional coordinates of the first frame after contact coincide with the three-dimensional coordinates of the second frame after contact, the ejection velocity vector is determined to be zero, indicating that the fragment failed to penetrate the casing.

[0033] If the launch velocity vector is non-zero, the direction angle of that vector is recorded as the direction of the remaining velocity vector. This direction angle includes the launch elevation angle relative to the radial surface of the casing and the launch azimuth angle relative to the circumferential direction of the casing.

[0034] Furthermore, S5 includes:

[0035] The method for identifying the pixel regions occupied by debris in images captured by a high-speed camera group is as follows:

[0036] Traverse all pixels in the image, compare the gray value of each pixel with a dynamic threshold, and classify pixels with gray values ​​greater than the dynamic threshold as bright areas. The connected region with the largest area in the bright area is determined as the fragment region. Calculate the average coordinate of all pixels in the fragment region as the two-dimensional pixel coordinate of the fragment in the image, and record the pixel area of ​​the fragment region. If the pixel area changes by more than twice between two consecutive frames, it is determined that the fragment broke or collided with the inner wall of the casing at that frame.

[0037] For the same sampling time, multiple two-dimensional pixel coordinates are obtained. Using a pre-calibrated projection matrix, these coordinates are converted into spatial rays emanating from the viewpoint of the camera group. The coordinates of the spatial point with the minimum sum of squared distances between all rays are then calculated. The calculation method is as follows:

[0038] The ray equation corresponding to each camera group is written in the form of the distance from a point to a line. An objective function is established, and the spatial coordinates that minimize the objective function are found through numerical iteration. The spatial coordinates of the spatial point are used as the three-dimensional spatial coordinates of the fragment at that sampling time.

[0039] The three-dimensional spatial coordinates obtained from multiple consecutive sampling times are arranged to form a discrete spatial position sequence of fragments. For each intermediate point in the sequence, the displacement vector between it and the previous point and the average velocity vector corresponding to the displacement vector are calculated. Then, the average velocity vector corresponding to the displacement vector between it and the next point is calculated. The average value of these two average velocity vectors is taken as the instantaneous velocity vector estimate at that point. The end points of all instantaneous velocity vectors are connected to form a velocity vector chain.

[0040] For the deformation boundary line of the inner wall of the casing, extract the intersection points of the pre-made grid marking lines on the surface of the inner wall of the casing. The method for extracting the intersection points is as follows:

[0041] Search for the continuous pixel line with the largest gray-level gradient along the row and column directions in the image, and determine the intersection point as the intersection point to obtain the two-dimensional pixel coordinates of each intersection point.

[0042] The two-dimensional coordinates of the intersection point are converted into three-dimensional spatial coordinates using a projection matrix. After the consistency check is performed on the three-dimensional spatial coordinates obtained by all camera groups, the average value is taken as the final three-dimensional spatial coordinates of the intersection point.

[0043] Compare the three-dimensional spatial coordinates of the intersection of the same grid marking line before deformation with the three-dimensional spatial coordinates after deformation, calculate the displacement vector of each intersection, and the area enclosed by the intersections whose displacement vector magnitude exceeds the preset threshold is the edge of the impact depression area.

[0044] The discrete spatial position sequence of the fragments, the velocity vector chain, and the closed polygon of the impact depression region boundary are stored together in the data recording unit of the test bench as input data for subsequent judgment of fragment penetration behavior.

[0045] Furthermore, the rotor support structure described in S1 includes a detachable rotor simulation disk, with the central shaft supported on the base of the test bench by a magnetic levitation bearing. The stator part of the magnetic levitation bearing is fixed inside the base, and the rotor part is sleeved on the central shaft. During loading, the magnetic levitation bearing actively adjusts the radial position of the central shaft.

[0046] The outer edge of the rotor simulation disk is provided with multiple slots for mounting the rotor components to be tested. Each slot has a strain gauge at the bottom for measuring the preload. The opening direction of the slot is consistent with the radial direction of the rotor simulation disk.

[0047] When installing the rotor component to be tested, insert the root of the rotor component into the slot and lock it with a wedge block. The locking torque of the wedge block is adjusted to a preset force range according to the strain gauge reading. This force range corresponds to the equivalent value of the centrifugal force borne by the rotor component in the actual engine.

[0048] Furthermore, when extracting the deformation boundary line of the inner wall of the casing in S5, for the reference image before deformation and each frame image after deformation, the displacement field of each sub-region is calculated using a local sub-region matching method. The specific matching method is as follows:

[0049] The image is divided into multiple overlapping sub-regions, each of which is a square pixel block. The gray-level distribution of one sub-region in the reference image is selected as a template. The sub-region position with the highest correlation to the gray-level distribution of the template is searched in the deformed image. The displacement vector of the center of the sub-region is obtained by the offset of the position relative to the original position of the template.

[0050] Interpolate the displacement vectors of all sub-region centers to obtain a displacement field cloud map of the entire inner wall surface of the casing. Extract continuous pixel bands whose displacement gradient exceeds the gradient threshold from this displacement field cloud map as deformation boundary lines.

[0051] According to a second aspect of the present invention, the present invention claims protection for a high-speed camera monitoring system for electromagnetic dynamic dual-axis synchronous loading, comprising:

[0052] One or more processors;

[0053] A memory storing one or more programs, which, when executed by one or more processors, enable the one or more processors to implement the high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading.

[0054] This invention discloses a high-speed camera monitoring method and system for electromagnetic dynamic biaxial synchronous loading, belonging to the field of aerospace component performance testing technology. The turbine engine rotor component to be monitored is mounted on an electromagnetic dynamic biaxial loading test bench. Radial and axial electromagnetic loaders are driven to simulate alternating load conditions during actual engine operation. A fracture excitation device is activated to cause instantaneous fracture at a predetermined location, releasing fragments and acquiring a sequence of fragment flight images. The two-dimensional images are unified into three-dimensional space, and the three-dimensional flight trajectory coordinate sequence of the fragments and the deformation boundary line of the casing inner wall are extracted. The incident velocity vector of the fragment impacting the casing inner wall and the exit velocity vector after penetration are calculated to determine whether the fragment has penetrated the casing. This invention can realistically reproduce the fracture and dispersion process of high-speed, high-energy fragments under biaxial dynamic loading, improving the accuracy of three-dimensional trajectory measurement and the reliability of penetration determination, providing accurate experimental basis for the containment design and airworthiness verification of aero-engine casings. Attached Figure Description

[0055] Figure 1 A flowchart illustrating the workflow of a high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading, as claimed in an embodiment of the present invention.

[0056] Figure 2This is a second flowchart of a high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading, as claimed in an embodiment of the present invention.

[0057] Figure 3 The third flowchart is a high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading, as claimed in an embodiment of the present invention. Detailed Implementation

[0058] 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 a part of the embodiments of this application, and not all of the 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.

[0059] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of those features. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications in the embodiments of this application, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationships and movements between components in a specific orientation as shown in the accompanying drawings. If the specific orientation changes, the directional indications will change accordingly. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0060] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0061] According to the first embodiment of the present invention, referring to Figure 1 This invention claims protection for a high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading, comprising:

[0062] S1: The turbine engine rotor component to be monitored is installed on the rotor support structure of the electromagnetic dynamic biaxial loading test bench. Radial electromagnetic loaders are arranged in the radial direction and axial electromagnetic loaders are arranged in the axial direction of the test bench. Multiple high-speed camera groups with optical spindles pointing to the preset debris scattering space area are arranged around the outer casing.

[0063] S2: By sending loading commands to the radial electromagnetic loader and the axial electromagnetic loader respectively through an independent timing controller, the two electromagnetic loaders synchronously apply dynamically changing electromagnetic forces to generate a dual-axis synchronous preload on the rotor component that simulates the operating conditions of an engine.

[0064] S3: Activate the fracture excitation device installed at the predetermined fracture site of the rotor component, so that the predetermined fracture site will fracture instantaneously under the combined action of biaxial preload, releasing multiple fragments.

[0065] S4: The timing controller sends synchronous trigger pulses to all high-speed camera groups to acquire images from the moment the fragment breaks until the fragment hits the inner wall of the casing and stops moving, recording the instantaneous image of the fragment in space and the deformation stripe image of the inner wall surface of the casing.

[0066] S5: Extract the geometric centroid coordinates of the fragments and the deformation boundary line of the inner wall of the casing frame by frame from the image sequence, and use the pre-calibrated spatial coordinate transformation matrix to obtain the flight trajectory coordinate sequence of the fragments in three-dimensional space and the boundary of the impact indentation area of ​​the inner wall of the casing.

[0067] S6: Calculate the incident velocity vector and the exit velocity vector after penetration at the impact point of the fragment on the inner wall of the casing based on the flight trajectory coordinate sequence. Determine whether the fragment has penetrated the casing based on whether the exit velocity vector is zero, and record the direction of the remaining velocity vector when penetration occurs.

[0068] In this embodiment, the method was used to monitor the trajectory of debris and the containment capacity of the casing after blade fracture during a foreign object impact test on the rotor blades of a certain type of aero-turbine engine. The specific implementation process is as follows:

[0069] First, the turbine engine rotor component to be monitored—a third-stage rotor blade of a high-pressure compressor—is mounted on the rotor support structure of the electromagnetic dynamic biaxial loading test bench. The rotor support structure includes a rotatable disk simulator, and the blade is fixed in the corresponding slot on the edge of the disk by a dovetail tenon. Four radial electromagnetic loaders are arranged radially on the test rig, evenly spaced at 90-degree intervals along the circumference of the wheel. The core axis of each radial electromagnetic loader coincides with the radial direction of the wheel, and the end face of the core faces the middle of the blade. Two axial electromagnetic loaders are arranged axially on the test rig, located at the front and rear faces of the wheel, respectively. Each axial electromagnetic loader is a ring structure with an inner diameter larger than the diameter of the wheel shaft, and the ring end face faces the ring area of ​​the wheel end face. Eight high-speed camera groups are arranged around the outer casing of the test rig, with these camera groups arranged at equal angles along the circumference of the casing. Each camera group contains a high-speed camera and a corresponding fixed-focus lens. The optical axis of the lens is precisely adjusted to point to the preset debris scattering space area—that is, the entire conical space area from the blade breakage initiation position to the inner wall of the casing, ensuring that at least two camera groups can capture the debris simultaneously, regardless of the direction in which it flies out.

[0070] Next, loading commands are sent to the radial and axial electromagnetic loaders via independent timing controllers. Each timing controller has two independent output channels, each connected to a power amplifier, which in turn drives the excitation coils of the radial and axial electromagnetic loaders. The timing controller first outputs a slowly rising ramp voltage waveform to the radial loader, generating a gradually increasing radial electromagnetic force. This radial electromagnetic force acts on the blade to simulate the radial tensile load generated by centrifugal force. Simultaneously, the timing controller outputs a sinusoidal waveform synchronized with the radial waveform but with a phase delay of a preset angle to the axial loader, allowing the axial electromagnetic force to simulate the axial thrust load generated by airflow in the engine. The two electromagnetic loaders synchronously apply dynamically changing electromagnetic forces, establishing a dual-axis synchronous preload on the blade. The magnitude and phase relationship of this load are predetermined based on the stress distribution at the blade root under maximum takeoff thrust conditions for this type of engine. Throughout the loading process, strain gauges installed at the blade root monitor the actual load in real time, ensuring that the load level reaches the target value before fracture induction.

[0071] Then, the fracture excitation device installed at the predetermined fracture location of the rotor component is activated. In this embodiment, the predetermined fracture location is set at the transition arc where the blade body and the tenon are connected. A narrow groove with a depth of one-quarter of the blade thickness is pre-machined at this location. The fracture excitation device includes a miniature electric explosion wire embedded at the bottom of the narrow groove and a high-voltage pulse power supply. When the timing controller detects that the dual-axis load is stable at the target value and continues for a preset time, it sends a trigger signal to the high-voltage pulse power supply. The high-voltage pulse power supply instantly releases the stored energy onto the electric explosion wire, causing the electric explosion wire to vaporize and generate a shock wave, which causes the blade to break instantaneously along the narrow groove, releasing at least one fragment—that is, the main body of the rotor after the blade breaks off.

[0072] Simultaneously with the activation of the fracture excitation device, the timing controller sends synchronous trigger pulses to all high-speed camera groups. The timing controller uses a high-speed comparator circuit to monitor the discharge current of the fracture excitation device. When the current rise exceeds a threshold, a TTL-level synchronous trigger pulse is immediately generated. This pulse is fanned out and split into eight paths, which are then transmitted in parallel through equal-length RF coaxial cables to the dedicated trigger input interface of each high-speed camera group. Upon receiving the trigger pulse, each camera group continuously captures tens of thousands of frames per second at a pre-set frame rate, such as determined through pre-experiments, capturing a series of images from the moment the fragment breaks until it impacts the inner wall of the casing and stops moving. Each frame simultaneously records the instantaneous image of the fragment in space and the deformation stripe image on the surface of the inner wall of the casing—the inner wall of the casing is pre-coated with a high-contrast random speckle pattern, allowing the deformation to be clearly distinguished.

[0073] After image acquisition, the geometric centroid coordinates of the fragments and the deformation boundary lines of the casing inner wall are extracted frame by frame from the image sequences acquired by each high-speed camera group. For each frame, the connected component labeling algorithm is first used to find the bright area corresponding to the fragment, and the average value of all pixel coordinates in the area is calculated as the centroid coordinates. For the deformation boundary lines, the displacement of each sub-region in the speckle pattern of the casing inner wall is calculated using digital image correlation methods, and then the continuous pixel bands where the displacement gradient changes abruptly are identified as the boundary lines. Using the projection matrix corresponding to each camera group obtained by calibrating the target before the experiment, the centroid coordinates extracted from the images taken by different camera groups at the same time are unified and converted into three-dimensional points in the same world coordinate system. The three-dimensional points at all times are arranged in chronological order to obtain the flight trajectory coordinate sequence of the fragments in three-dimensional space. At the same time, the three-dimensional coordinates of the deformation boundary lines at each time are combined to obtain the spatiotemporal evolution data of the impact depression region boundary.

[0074] Finally, based on the flight trajectory coordinate sequence, the incident velocity vector at the impact point on the inner wall of the casing and the exit velocity vector after penetration are calculated. The incident velocity vector is obtained by dividing the difference between the three-dimensional coordinates of the frame before the impact and the three-dimensional coordinates of the frame before that by the inter-frame time difference. Similarly, the exit velocity vector is calculated by selecting the three-dimensional coordinates of the first and second frames after the impact. If all three components of the exit velocity vector are zero, that is, the position of the fragment does not change in two consecutive frames after the impact and is located outside the outer wall of the casing, it is determined that the fragment has penetrated the casing. The position of the fragment at the time of penetration and the velocity direction in the last frame before the exit velocity vector is zero are recorded as the direction of the remaining velocity vector. If the exit velocity vector is not zero, but the magnitude is less than a certain proportion of the magnitude of the incident velocity vector, it is also determined that the fragment has penetrated but most of the energy has been absorbed.

[0075] Furthermore, the radial electromagnetic loader described in S1 includes a plurality of radial electromagnets evenly distributed along the circumference of the rotor component, the core axis of each radial electromagnet coincides with the radial direction of the rotor component, and a non-contact air gap is maintained between the core end face of each radial electromagnet and the outer edge surface of the rotor component.

[0076] The axial electromagnetic loader includes at least two coaxially arranged annular electromagnets, which are respectively disposed on the front end face and the rear end face of the rotor component. The inner diameter of each annular electromagnet is larger than the diameter of the rotor shaft of the rotor component. The end face of the annular electromagnet and the corresponding end face of the rotor component also maintain a non-contact air gap.

[0077] The inner wall surface of the outer casing is coated with a diffuse reflection coating, which is formed by spraying a mixture of magnesium oxide powder and epoxy resin in a certain proportion, and is used to enhance the texture contrast of the inner wall of the casing in high-speed camera images.

[0078] In this embodiment, the radial electromagnetic loader employs four C-shaped radial electromagnets. The core of each electromagnet is made of stacked silicon steel sheets, and the end of the core is machined into an arc-shaped surface that matches the outer edge profile of the blade. These four radial electromagnets are evenly distributed circumferentially along the rotor component, and the core axis of each electromagnet coincides with the radial direction of the rotor component, i.e., pointing outward from the center of the test bench. A non-contact air gap is maintained between the arc-shaped end face of the core and the outer edge surface of the rotor blade. This air gap is adjusted by a precision micro-displacement adjustment mechanism mounted on the core support. During adjustment, a feeler gauge is used to measure the air gap size, ensuring that the air gap maintains a preset gap value when the blade is stationary. This gap value must ensure effective transmission of electromagnetic force while preventing collision between the blade and the core during high-speed rotation or vibration. The excitation coil of each radial electromagnet is powered by an independent power amplifier to independently control the electromagnetic force component in each direction.

[0079] For the axial electromagnetic loader, two coaxially arranged annular electromagnets are used, fixed to the front and rear plates of the test bench, respectively. The annular electromagnet on the front plate faces the front face of the rotor disk, and the annular electromagnet on the rear plate faces the rear face of the disk. The core of each annular electromagnet is also made of silicon steel sheets, and its inner diameter is larger than the diameter of the rotor shaft by at least a preset margin to ensure that the shaft will not touch the inner wall of the annular electromagnet when it undergoes radial displacement under the support of the magnetic levitation bearing. The end face of the annular electromagnet is machined into a flat surface, maintaining a non-contact air gap with the corresponding end face of the disk. The air gap is adjusted in a similar way to that of the radial electromagnet. The excitation coils of the two annular electromagnets can be driven in series or in parallel, or independently, depending on the axial force distribution pattern to be simulated. When it is necessary to simulate uneven axial thrust, the front and rear annular electromagnets apply electromagnetic forces of different magnitudes, thereby generating additional bending moment loads on the rotor.

[0080] The inner wall surface of the outer casing is coated with a diffuse reflection coating. This coating is formed by mixing magnesium oxide powder and epoxy resin in a specific ratio: first, analytically pure magnesium oxide powder is dried in an oven to remove moisture, then mixed with bisphenol A type epoxy resin and a curing agent at a volume ratio of approximately 1:2, and stirred until a paste is formed. This paste is then evenly sprayed onto the inner wall surface of the casing using a high-pressure airless spray gun, with the spray thickness controlled to completely cover the metal substrate without sagging. After spraying, it is cured at room temperature. The coating has a white matte finish. The coating has a much higher diffuse reflectance of visible light than that of a bare metal surface. When the strobe illumination of a high-speed camera shines on the inner wall of the casing, the coating scatters the light evenly in all directions, making the metal surface, which is prone to specular reflection, uniformly bright and uniform, thus forming high-contrast texture features in the image. Even if the inner wall of the casing is dented or bulged by the impact of debris, the coating will remain attached as the substrate deforms, and will not produce excessive specular reflection highlights due to changes in the surface tilt angle. This greatly improves the reliability of deformation boundary line extraction in subsequent image processing.

[0081] Furthermore, the dynamically changing electromagnetic force described in S2 is a time-varying force in the form of a sine wave. The electromagnetic forces applied by the radial electromagnetic loader and the axial electromagnetic loader have the same oscillation frequency, and the phase difference between the electromagnetic forces in the two directions is preset according to the force phase relationship of the rotor components in actual engine operation.

[0082] The timing controller contains two independent waveform generation channels. Each channel outputs a programmable voltage waveform to the drive coil of the corresponding electromagnetic loader. The two channels are locked in phase by a clock synchronization signal, so that the radial electromagnetic force and the axial electromagnetic force can change in tandem according to a preset phase difference. During the application of the dual-axis synchronous preload, the magnitude of the electromagnetic force gradually increases from zero to the target amplitude. The rate of increase is determined according to the material strength and fracture toughness of the rotor components to avoid unexpected premature fracture before the fracture excitation device is activated.

[0083] In this embodiment, the electromagnetic forces applied by the radial and axial electromagnetic loaders are time-varying forces in the form of sinusoidal waves. The oscillation frequency of the sinusoidal waves is set according to the main frequency of the alternating load on the rotor blades in the actual engine, such as corresponding to the rotor rotation fundamental frequency or the blade passing frequency. The electromagnetic forces in the radial and axial directions have the same oscillation frequency to ensure the periodic synchronization of the loading. The phase difference between the electromagnetic forces in the two directions is preset according to the force phase relationship of the rotor components in the actual engine operation: in the high-pressure compressor of a certain type of engine, there is a fixed phase relationship between the radial centrifugal force and the axial aerodynamic thrust borne by the blades, because the axial thrust pulsation and the radial centrifugal force pulsation both originate from the same rotor rotation cycle. By adjusting the phase difference, it is possible to simulate, for example, the condition where the maximum radial force occurs when the axial force reaches a certain percentage during each rotor revolution.

[0084] The timing controller contains two independent waveform generation channels, each generating a programmable voltage waveform based on direct digital frequency synthesis technology. The two channels share a highly stable clock source, which outputs a reference frequency signal to the phase-locked loop circuits of both channels. Each channel contains a digital-to-analog converter (DAC) that converts the digital waveform sequence in the waveform memory into an analog voltage. This analog voltage is then amplified by a linear power amplifier and output to the drive coil of the corresponding electromagnetic loader. The two channels are synchronized by a clock signal to lock their phase relationship: one channel is designated as the master channel, and its output periodic synchronization pulse serves as the phase reset signal for the slave channel, ensuring that the waveform start point of the slave channel always maintains a preset phase offset from the waveform start point of the master channel. In this way, the radial and axial electromagnetic forces can coordinately change according to a preset phase difference, for example, the radial force leading the axial force by a preset angle.

[0085] During the application of biaxial synchronous preload, the magnitude of the electromagnetic force is not abrupt, but rather starts from an initial value close to zero and gradually increases to the target amplitude according to a slowly rising ramp curve. The rate of ramp increase is determined based on the material strength and fracture toughness of the rotor components. Specifically, before the test begins, the yield strength and fracture toughness values ​​of the rotor blade material are obtained by consulting the material handbook. Then, the rate of stress increase inside the blade under loading at this rate is calculated through finite element simulation to ensure that the stress level does not exceed a certain safe proportion of the material's yield strength at any time. This avoids premature fracture or plastic yielding of the blade due to excessively rapid load increase before the fracture excitation device is activated. During actual loading, the timing controller outputs a linear ramp envelope signal to modulate the amplitude of the sine wave, causing the peak value of the sine wave to increase linearly with time. The entire rising process lasts for a preset time until the target amplitude is reached and then remains constant, waiting for the fracture excitation signal.

[0086] Furthermore, referring to Figure 2 S4 specifically includes the following sub-steps:

[0087] S41: Generates rising edge synchronous trigger pulses inside the timing controller and transmits them in parallel to the external trigger interface of each high-speed camera group via coaxial cable;

[0088] S42: Each high-speed camera group is equipped with a high-precision crystal oscillator. When the external trigger interface receives the rising edge of the synchronous trigger pulse, the crystal oscillator starts counting and starts the first frame exposure of the camera group after reaching the preset delay count. The delay count is independently calibrated according to the optical path distance between each camera group and the fracture excitation position.

[0089] S43: Each high-speed camera group continuously performs exposure and readout operations for subsequent frames according to a pre-set fixed frame rate. The interval between the exposure start times of two adjacent frames is obtained by frequency division of the crystal oscillator. The fixed frame rate is determined based on the expected maximum speed of the debris flying in the internal space of the housing and the maximum allowable motion blur displacement of the debris in each frame image.

[0090] S44: A narrow-band filter is installed in front of the photosensitive surface of the image sensor of each high-speed camera group. The center wavelength is matched with the emission wavelength of the stroboscopic illumination source arranged around the test bench. During the exposure of each frame image, the timing controller sends a pulse drive current synchronized with the exposure window of the camera group to the stroboscopic illumination source. The edge contour of the fragment and the deformation characteristics of the inner wall surface of the casing are recorded.

[0091] S45: Each high-speed camera group continuously acquires images until it detects that the change in the position of a fragment in multiple consecutive frames is less than the spatial distance corresponding to one pixel. It then automatically stops acquiring images and stores the acquired complete image sequence in the non-volatile memory inside the high-speed camera group.

[0092] In this embodiment, simultaneously with receiving the start signal from the fracture excitation device, the timing controller internally generates a rising-edge synchronous trigger pulse. Specifically, a current transformer is connected in series in the high-voltage discharge circuit of the fracture excitation device, and the output of the transformer is connected to a high-speed comparator inside the timing controller. When the discharge current rises sharply, the voltage induced by the transformer exceeds the preset threshold of the comparator, causing the comparator output to flip and triggering a monostable multivibrator to generate a TTL positive pulse with a width on the order of microseconds. This pulse is driven by a high-speed buffer and then sent to a coaxial cable distributor with one input and eight outputs. Each output port of the distributor is connected to an RG-58 coaxial cable of the same physical length. The length difference of all cables is controlled within the range that the time delay difference caused by the propagation distance of electromagnetic waves in the cable is less than the order of magnitude of the minimum exposure time of the high-speed camera group. For example, if the minimum exposure time of the camera is 1 microsecond, the cable length difference is controlled within 1 microsecond corresponding to 300 meters. In practice, the cable length difference does not exceed 0.3 meters, and the corresponding time delay difference is less than 1 nanosecond.

[0093] Each high-speed camera group is equipped with a high-precision crystal oscillator with a frequency stability better than one part per million. When the external trigger interface receives the rising edge of the synchronous trigger pulse, the crystal oscillator starts counting. Each camera group's programmable logic device has a delay counter with a preset delay count value. When the counter reaches the preset value from zero, an internal trigger signal is generated to start the first frame exposure of that camera group. The delay count is independently calibrated based on the optical path distance between each camera group and the fracture excitation position: the spatial straight-line distance between the fracture excitation position, i.e., the predetermined fracture part of the blade, and the front end of each camera lens is different, and the time required for light to travel from the fracture position to the sensors of different cameras is also different. In order to keep the time difference between the moment when all cameras capture the first frame image and the moment when the fragment leaves the fracture position consistent, calibration is required. Calibration method: A miniature LED instantaneous light source is placed at the fracture excitation location. This light source is driven by a separate output pulse from a timing controller, which has a fixed time relationship with the synchronous trigger pulse. All camera groups simultaneously capture the lighting process of this instantaneous light source. By comparing the frame number of the first appearance of the light source in the actual image sequence recorded by each camera group and the pixel position within that frame, the delay deviation of each camera group relative to the reference camera group can be calculated. Then, the delay count value of each camera group is adjusted so that in the formal test, when the fragment arrives at the center of the field of view of each camera after leaving the fracture location and taking the same optical path time, all cameras have just begun to expose the first frame.

[0094] Each high-speed camera group continuously performs exposure and readout operations for subsequent frames according to a pre-set fixed frame rate; the interval between the exposure start times of two adjacent frames is obtained by frequency division of the crystal oscillator. Specifically, the output of the crystal oscillator passes through a programmable frequency divider, and the division ratio is set by the user according to the desired frame rate; the fixed frame rate is determined based on the expected maximum speed of the fragment flying in the internal space of the housing and the maximum allowable motion blur displacement of the fragment in each frame image; to avoid blurring, the distance the fragment moves within a single frame exposure time is required not to exceed the spatial dimension corresponding to one pixel; through preliminary experiments, using simulated projectiles of the same material and size as real fragments, fired in a housing model of the same size, the frame rate of the high-speed camera is gradually increased, starting from tens of thousands of frames per second. Each time the frame rate is increased, the clarity of the edges of the simulated projectiles in the captured images is observed; when the edge clarity reaches a point where the machining marks on the surface of the projectile can be distinguished, the frame rate is considered qualified, and the frame rate is fixed for formal testing.

[0095] Each high-speed camera array has a narrowband filter installed in front of its image sensor's photosensitive surface. This narrowband filter has a center wavelength of, for example, 532 nm and a half-width at half-maximum (FWHM) of 10 nm. Simultaneously, high-power LED arrays are arranged at multiple locations around the test bench as strobe illumination sources; these LEDs emit at a wavelength of 532 nm. During the exposure of each frame, a timing controller sends a pulsed drive current synchronized with the camera array's exposure window to the strobe illumination sources. The width of this pulsed drive current is set to half the exposure time of a single frame for the camera array. Internally, the timing controller processes the exposure start signal for each frame through a... A monostable multivibrator generates a pulse with adjustable width, which drives an LED array via a power MOSFET. The LED array emits high-intensity short pulses of light during the pulse and is turned off in the latter half of the exposure window. Since the LED pulse width is less than half the exposure time, the image sensor mainly receives ambient light, which is already very weak after passing through a narrow-band filter, in the first half of the exposure window, while receiving LED pulse light in the latter half. Because the intensity of the pulse light is much higher than that of the ambient light, the final recorded image is equivalent to being exposed only during the pulse illumination period, thus effectively freezing the high-speed movement of debris and avoiding motion blur caused by continuous illumination.

[0096] Each high-speed camera group continuously acquires images until it detects that the change in the position of a fragment in multiple consecutive frames is less than the spatial distance corresponding to one pixel, at which point it automatically stops acquiring images. During detection, the internal processor of each camera group analyzes the centroid coordinates of the fragment in the last two frames in real time and calculates the pixel displacement. If the displacement is less than one pixel in three consecutive frames, it is determined that the fragment has stopped moving, and the processor sends an acquisition end flag to the timing controller, which then cuts off the trigger signal of the camera. The acquired complete image sequence is stored in the non-volatile memory inside the high-speed camera group, which is a high-speed solid-state drive array. The timestamp corresponding to the acquisition time of each frame is converted from the count value of the crystal oscillator—that is, the crystal oscillator starts counting after the arrival of the synchronization trigger pulse. The counter value at the start of exposure for each frame is recorded and saved as a timestamp, and the timestamp and image data are stored in the same data packet in a one-to-one correspondence.

[0097] Furthermore, the pre-calibrated spatial coordinate transformation matrix described in S5 is obtained in the following way:

[0098] A calibration target is placed in the debris scattering area, and multiple high-speed camera groups simultaneously capture images of the calibration target to extract the two-dimensional pixel coordinates of the spherical marker points.

[0099] For each camera group, establish the projection relationship between the two-dimensional pixel coordinates extracted from the image of the group and the three-dimensional coordinates of the corresponding marker points on the calibration target, and solve for the projection matrix corresponding to the group.

[0100] By combining the projection matrices of all camera groups into a set of three-dimensional spatial coordinate transformation equations, the transformation matrix from the two-dimensional pixel coordinates of any camera group to the unified three-dimensional spatial coordinates can be obtained.

[0101] In this embodiment, a calibration target is placed within the debris scattering area. The calibration target is a cubic frame, the side length of which is designed according to the internal space dimensions of the casing to cover the entire space range where debris may appear. Each vertex and the midpoint of each edge of the frame is inlaid with a spherical marker of uniform diameter. These spherical markers are made of white ceramic material with a diffuse reflection treatment. Using a high-precision coordinate measuring machine, under stable laboratory temperature conditions, the three-dimensional coordinates of the center of each spherical marker are measured one by one with a measurement accuracy of micrometer level, and these coordinate values ​​are recorded as a target spatial point set.

[0102] Multiple high-speed camera groups simultaneously captured images of the calibration target. During shooting, the focal length, aperture, and focus position of all cameras were kept fixed and identical to those used in the actual experiment. The two-dimensional pixel coordinates of the spherical marker points were extracted from the images captured by each camera group. The extraction method was as follows: First, Gaussian filtering was applied to the images for noise reduction. Then, the Canny edge detection operator was used to extract the elliptical edges projected from each spherical marker. Finally, least-squares ellipse fitting was applied to the edge point set to obtain the coordinates of the ellipse's center, which were then used as the two-dimensional pixel coordinates of the spherical marker point. Since the projection of the marker point's center in space does not strictly coincide with the ellipse's center, perspective projection distortion exists, but ellipse fitting can approximate this distortion well.

[0103] For each camera group, a projection relationship is established between the 2D pixel coordinates extracted from the image of that group and the 3D coordinates of the corresponding marker points on the calibration target. The projection relationship adopts a pinhole camera model, including the intrinsic parameter matrix (focal length, principal point coordinates, distortion coefficients) and the extrinsic parameter matrix (rotation matrix and translation vector). When solving for the projection matrix of each group, the idea of ​​Zhang Zhengyou's calibration method is used. The 3D coordinates of all marker points and their corresponding 2D pixel coordinates are substituted into the model, and the optimal projection matrix parameters are iteratively solved by minimizing the reprojection error.

[0104] The projection matrices of all camera groups are combined to form a set of three-dimensional spatial coordinate transformation equations. Specifically, for any target point in space, it is imaged in the images of two different camera groups. Based on the projection matrix of each camera, two spatial ray equations can be listed. The ray equations of multiple cameras are combined, and an overdetermined set of equations is obtained using linear triangulation. Solving the least squares solution of this set of equations yields the transformation matrix from the two-dimensional pixel coordinates of any camera group or the coordinates of multiple camera groups to a unified three-dimensional spatial coordinate system. This transformation matrix is ​​essentially an operator that maps multiple two-dimensional observations to three-dimensional points. The optimization objective during its solution is to minimize the sum of squared reprojection errors of all marked points. During optimization, the known three-dimensional coordinates of all marked points are reprojected into each image using the currently estimated projection matrix. The sum of squared Euclidean distances between the reprojected coordinates and the actual extracted ellipse center coordinates is calculated. Then, the Levenberg-Marquardt algorithm is used to continuously adjust the projection matrix parameters until the sum of squared errors converges to a minimum value. The final transformation matrix and the projection matrices of each camera group are saved for use by S5.

[0105] Furthermore, the incident velocity vector described in S6 is calculated in the following manner:

[0106] The three-dimensional coordinates of the frame before the fragment contacts the inner wall of the casing and the three-dimensional coordinates of the frame before the contact are selected from the fragment flight trajectory coordinate sequence. The difference vector is divided by the time interval between the two frames to obtain the incident velocity vector.

[0107] The ejection velocity vector is calculated in the following way:

[0108] Select the three-dimensional coordinates of the first frame after the fragment contacts the inner wall of the casing and the three-dimensional coordinates of the second frame after contact from the fragment flight trajectory coordinate sequence. Divide the difference vector by the time interval between the two frames to obtain the ejection velocity vector. If the three-dimensional coordinates of the first frame after contact coincide with the three-dimensional coordinates of the second frame after contact, the ejection velocity vector is determined to be zero, indicating that the fragment failed to penetrate the casing.

[0109] If the launch velocity vector is non-zero, the direction angle of that vector is recorded as the direction of the remaining velocity vector. This direction angle includes the launch elevation angle relative to the radial surface of the casing and the launch azimuth angle relative to the circumferential direction of the casing.

[0110] In this embodiment, the three-dimensional coordinates of the frame before the fragment contacts the inner wall of the casing and the frame before the contact frame are selected from the fragment flight trajectory coordinate sequence obtained in S5. It is necessary to accurately determine the frame number where the contact occurs. The determination criteria are as follows: traverse the entire trajectory coordinate sequence and calculate the magnitude and direction of the displacement vector between adjacent points. Under normal circumstances, when the fragment flies in the air, the displacement remains basically constant frame by frame, ignoring air resistance, and the direction is also basically consistent. When the fragment hits the inner wall of the casing, the collision causes a sharp change in velocity direction and a sudden decrease in velocity magnitude. Therefore, the displacement vector direction between adjacent points will change abruptly, for example, the angle is greater than a preset angle threshold, such as 90 degrees, and the displacement decreases sharply to a proportion lower than a preset proportion threshold, such as less than 10%. When these two conditions are met, the intermediate frame between the last point before the collision of the previous point and the first point after the collision of the next point is determined as the contact frame. In actual implementation, the frame before the contact frame is taken as the contact frame, and the two frames before the contact frame are taken as the frame before the contact frame. The incident velocity vector is obtained by dividing the difference vector between the two three-dimensional coordinates by the time interval between the corresponding two frames and subtracting the timestamps; the direction of this vector points towards the inner wall of the casing.

[0111] Then, select the three-dimensional coordinates of the first frame after the debris contacts the inner wall of the casing and the three-dimensional coordinates of the second frame after contact from the debris flight trajectory coordinate sequence; similarly, the first frame after contact is the frame after the contact frame, and the second frame after contact is the frame after that; divide the difference vector of these two three-dimensional coordinates by the time interval between the corresponding two frames to obtain the launch velocity vector.

[0112] Penetration determination rules: If the three-dimensional coordinates of the first frame after contact coincide with the three-dimensional coordinates of the second frame after contact (i.e., the Euclidean distance between them is less than the spatial distance corresponding to one pixel), and the coincident position is located on the outer side of the casing wall, determined by the casing wall thickness and the normal direction of the impact point, then the ejection velocity vector is determined to be a zero vector, indicating that the fragment was completely stopped by the casing wall or external obstruction after penetrating the casing; if the two do not coincide, but the magnitude of the ejection velocity vector is less than a preset proportion of the magnitude of the incident velocity vector, such as one percent, and the direction of the ejection velocity vector is opposite to the direction of the incident velocity vector, i.e., it bounces back into the casing, then the fragment is also determined to have failed to penetrate the casing, and is a zero vector; if the ejection velocity vector is non-zero and its direction points to the outside of the casing, then the fragment is determined to have successfully penetrated, and the direction angle of the vector is recorded; the direction angle includes the ejection elevation angle relative to the radial surface of the casing and the ejection azimuth angle relative to the circumferential direction of the casing: taking the tangent plane at the impact point of the casing as a reference, the angle between the ejection velocity vector and the tangent plane is the elevation angle, and the angle between the projection on the tangent plane and the circumferential tangent is the azimuth angle.

[0113] Furthermore, S5 includes:

[0114] The method for identifying the pixel regions occupied by debris in images captured by a high-speed camera group is as follows:

[0115] Traverse all pixels in the image, compare the gray value of each pixel with a dynamic threshold, and classify pixels with gray values ​​greater than the dynamic threshold as bright areas. The connected region with the largest area in the bright area is determined as the fragment region. Calculate the average coordinate of all pixels in the fragment region as the two-dimensional pixel coordinate of the fragment in the image, and record the pixel area of ​​the fragment region. If the pixel area changes by more than twice between two consecutive frames, it is determined that the fragment broke or collided with the inner wall of the casing at that frame.

[0116] For the same sampling time, multiple two-dimensional pixel coordinates are obtained. Using a pre-calibrated projection matrix, these coordinates are converted into spatial rays emanating from the viewpoint of the camera group. The coordinates of the spatial point with the minimum sum of squared distances between all rays are then calculated. The calculation method is as follows:

[0117] The ray equation corresponding to each camera group is written in the form of the distance from a point to a line. An objective function is established, and the spatial coordinates that minimize the objective function are found through numerical iteration. The spatial coordinates of the spatial point are used as the three-dimensional spatial coordinates of the fragment at that sampling time.

[0118] The three-dimensional spatial coordinates obtained from multiple consecutive sampling times are arranged to form a discrete spatial position sequence of fragments. For each intermediate point in the sequence, the displacement vector between it and the previous point and the average velocity vector corresponding to the displacement vector are calculated. Then, the average velocity vector corresponding to the displacement vector between it and the next point is calculated. The average value of these two average velocity vectors is taken as the instantaneous velocity vector estimate at that point. The end points of all instantaneous velocity vectors are connected to form a velocity vector chain.

[0119] Reference Figure 3 For the deformation boundary line of the inner wall of the casing, the intersection points of the pre-made grid marking lines on the surface of the inner wall of the casing are extracted. The method for extracting the intersection points is as follows:

[0120] Search for the continuous pixel line with the largest gray-level gradient along the row and column directions in the image, and determine the intersection point as the intersection point to obtain the two-dimensional pixel coordinates of each intersection point.

[0121] The two-dimensional coordinates of the intersection point are converted into three-dimensional spatial coordinates using a projection matrix. After the consistency check is performed on the three-dimensional spatial coordinates obtained by all camera groups, the average value is taken as the final three-dimensional spatial coordinates of the intersection point.

[0122] Compare the three-dimensional spatial coordinates of the intersection of the same grid marking line before deformation with the three-dimensional spatial coordinates after deformation, calculate the displacement vector of each intersection, and the area enclosed by the intersections whose displacement vector magnitude exceeds the preset threshold is the edge of the impact depression area.

[0123] The discrete spatial position sequence of the fragments, the velocity vector chain, and the closed polygon of the impact depression region boundary are stored together in the data recording unit of the test bench as input data for subsequent judgment of fragment penetration behavior.

[0124] In this embodiment, for each frame of image captured by each high-speed camera group, the pixel region occupied by the fragment is identified in the image; the grayscale data of the frame image is read, all pixels are traversed, and the arithmetic mean of the grayscale values ​​of all pixels in the entire image is calculated, and this average value is used as a dynamic threshold; then the image is traversed again, and pixels with grayscale values ​​greater than the dynamic threshold are marked as bright pixels, and the rest are marked as background; in the set of bright pixels, an eight-connected region labeling algorithm is used to find all connected regions, and the pixel area of ​​each region is calculated; since fragments usually have large geometric dimensions and strong surface reflection, they will form a large connected region in the image with an approximately elliptical or elongated shape; while isolated bright spots caused by noise points or uneven coating have very small areas; therefore, the connected region with the largest area is determined to be the fragment region; the average row coordinates and average column coordinates of all pixels in the fragment region are calculated as the two-dimensional pixel coordinates of the fragment in the image; at the same time, the pixel area of ​​the fragment region is recorded. If the area of ​​a pixel changes by more than double between two consecutive frames, for example, from 1000 pixels to more than 2500 pixels, or from 1000 pixels to less than 400 pixels, it is determined that the fragment has broken at that frame. The area suddenly increases because a fragment has split into multiple fragments that overlap in the image projection or collide with the inner wall of the casing. The area suddenly decreases because the fragment is deformed or partially obscured. This determination result is stored in the data record as auxiliary information.

[0125] Then, for the same sampling time, i.e. the same frame number, multiple two-dimensional pixel coordinates are obtained from all high-speed camera groups; each camera group corresponds to one pixel coordinate; using the pre-calibrated projection matrix of the camera group, the two-dimensional pixel coordinates are combined with the unknown depth value to obtain a spatial ray from the perspective of the camera group—this ray starts from the optical center of the camera, passes through the pixel point on the image, and extends to infinity; theoretically, the spatial rays corresponding to all camera groups should intersect at the true three-dimensional position point of the fragment; however, due to measurement and calibration errors, the rays do not strictly intersect; at this time, the coordinates of the spatial point with the minimum sum of squared distances between all rays are calculated.

[0126] The calculation method is as follows: Each ray is represented as a parametric equation, i.e., the optical center point plus the direction vector multiplied by the parameter; the distance from any point in space to a ray is equal to the length of the component of the line connecting that point and the optical center perpendicular to the direction vector; an objective function is established, which is equal to the sum of the squares of the distances from the coordinates of a point in space to all rays; a numerical iterative method, such as the Newton-Raphson method, is used to gradually adjust the position of the initial guess point, such as the average of the coordinates of all optical centers, so that the value of the objective function decreases; the iteration stops when the change in the objective function is less than a preset small threshold, and the coordinates of the point at this time are the three-dimensional spatial coordinates of the fragment at that sampling time.

[0127] Next, the three-dimensional spatial coordinates obtained from multiple consecutive sampling moments are arranged in chronological order to form a discrete spatial position sequence of the debris. For each intermediate point in the sequence, i.e., each point other than the first and last points, the displacement vector between it and the previous point and the average velocity vector obtained by dividing the displacement vector by the time interval are calculated. Then, the average velocity vector corresponding to the displacement vector between the next point and the previous point is calculated. The average of these two average velocity vectors is taken as the instantaneous velocity vector estimate at that point. For example, if the average velocity between the i-th point and the previous point is V(i-1→i), and the average velocity between the i-th point and the next point is V(i→i+1), then the instantaneous velocity vector estimate at that point is equal to (V(i-1→i)+V(i→i+1)) / 2. The terminal points of all instantaneous velocity vectors are connected in chronological order to form a velocity vector chain. Each node of the chain corresponds to the terminal position of the velocity vector at a certain moment. By observing the direction of the chain, the continuous changing trend of the velocity direction and magnitude of the debris during flight can be intuitively seen—for example, the change in the curvature of the chain corresponds to the acceleration direction.

[0128] Next, for the deformation boundary line of the inner wall of the casing, the intersection points of pre-made grid marking lines on the surface of the inner wall of the casing are extracted from the images acquired by each high-speed camera group. These grid marking lines are orthogonal grid lines drawn on the surface of the inner wall of the casing using a black high-temperature resistant coating, such as silicone heat-resistant paint. Before drawing, a baseline is projected using a laser level, and then the grid is sprayed using a template. The spacing of the grid lines remains constant in both the circumferential and axial directions of the inner wall of the casing, for example, one line at fixed intervals. The method for extracting the intersection points is as follows: scanning along the row direction in the image, detecting edge points where the grayscale value drops sharply from the white coating to the black line, and connecting these edge points to form horizontal lines; The sample is scanned along the column direction to form vertical lines; the intersection of the horizontal and vertical lines is determined as the intersection point; after obtaining the two-dimensional pixel coordinates of each intersection point, they are converted into three-dimensional spatial coordinates using the projection matrix of the camera group; the three-dimensional spatial coordinates converted by all camera groups are first checked for consistency: the Euclidean distance between the three-dimensional spatial coordinates of the same intersection point obtained by any two camera groups is calculated. If the distance is less than a preset tolerance, such as one percent of the casing wall thickness, it is retained; otherwise, it is considered that the identification of the intersection point in the frame is incorrect, and the data is discarded; the average value of the coordinates that pass the test is taken as the final three-dimensional spatial coordinates of the intersection point.

[0129] The 3D spatial coordinates of the intersection points of the same grid marking lines in the last frame before deformation (i.e., before fracture excitation) are compared with the 3D spatial coordinates of each frame after deformation (i.e., after fragment impact). The displacement vector of each intersection point is calculated, which is the post-deformation coordinate minus the pre-deformation coordinate. If the magnitude of the displacement vector exceeds a preset threshold (which is the elastic deformation limit converted from the yield strength of the casing inner wall material), the intersection point is considered to have entered the plastic deformation region. All intersection points exceeding the threshold are extracted, and the area they enclose is the boundary of the impact indentation region. The boundary fitting method is as follows: these intersection points are sorted according to their original grid topology connection relationship on the casing inner wall surface. If there are voids, the convex hull algorithm is used to obtain the minimum convex polygon. If the boundary is not closed, it is closed by spline interpolation. Finally, a closed polygon is formed, representing the boundary contour of the casing inner wall indentation region.

[0130] Finally, the discrete spatial position sequence of the fragments, the velocity vector chain, and the closed polygon of the impact depression region boundary are stored together in the data recording unit of the test bench. This unit is the hard drive of an industrial computer, and the data is saved in binary file format for direct retrieval when judging the fragment penetration behavior later.

[0131] Furthermore, the rotor support structure described in S1 includes a detachable rotor simulation disk, with the central shaft supported on the base of the test bench by a magnetic levitation bearing. The stator part of the magnetic levitation bearing is fixed inside the base, and the rotor part is sleeved on the central shaft. During loading, the magnetic levitation bearing actively adjusts the radial position of the central shaft.

[0132] The outer edge of the rotor simulation disk is provided with multiple slots for mounting the rotor components to be tested. Each slot has a strain gauge at the bottom for measuring the preload. The opening direction of the slot is consistent with the radial direction of the rotor simulation disk.

[0133] When installing the rotor component to be tested, insert the root of the rotor component into the slot and lock it with a wedge block. The locking torque of the wedge block is adjusted to a preset force range according to the strain gauge reading. This force range corresponds to the equivalent value of the centrifugal force borne by the rotor component in the actual engine.

[0134] In this embodiment, the rotor support structure includes a detachable rotor simulation disk made of high-strength alloy steel, with a disc-shaped surface and multiple weight-reduction holes. The central shaft of the rotor simulation disk is supported on the base of the test bench by magnetic levitation bearings. The magnetic levitation bearings are a five-degree-of-freedom active magnetic levitation bearing system, including radial and axial magnetic levitation bearings. The stator is fixed to a cylindrical mounting base inside the base, with stator windings arranged radially and axially respectively. The rotor is fitted onto the central shaft, and its surface is covered with a magnetic ring. During loading, the controller of the magnetic levitation bearing actively adjusts the current in the electromagnetic coil based on feedback signals from the displacement sensor on the central shaft, ensuring that the central shaft remains within a preset levitation gap range, such as the center of the gap. This eliminates mechanical contact friction, allowing the rotor simulation disk to rotate freely with extremely low damping or remain stationary.

[0135] The outer edge of the rotor simulation disk has multiple slots for mounting the rotor components under test. The number of slots can be six, eight, or more, depending on the test requirements. Each slot has a shallow groove at the bottom, and a resistance strain gauge is attached inside the groove. The strain gauge leads are led along the groove on the surface of the simulation disk to a slip ring or telemetry system to measure the preload applied by the slot to the blade root. The opening direction of the slot is consistent with the radial direction of the rotor simulation disk, that is, the slot extends radially inward from the edge of the disk. The cross-sectional shape of each slot matches the root shape of the rotor component under test. For example, for a dovetail blade root, the slot is machined into a corresponding dovetail groove.

[0136] When installing the rotor component to be tested, insert the root of the rotor component into the slot, and then insert a wedge block from the radially inner side of the slot; the shape of the wedge block matches the inclined surface on the inner side of the slot; use a torque wrench to tighten the locking screw behind the wedge block, causing the wedge block to move outward from the slot, thereby pressing the blade root. During the tightening process, observe the strain gauge readings. When the voltage value output by the strain gauge reaches the voltage corresponding to a preset force range, stop tightening; this force range is obtained through finite element calculations and corresponds to the equivalent value of the centrifugal force borne by the blade in an actual engine at rated speed. For example, if the centrifugal force of the blade in an actual engine is a certain value, the preload is adjusted so that the compressive stress on the blade root is the same as the compressive stress generated by the centrifugal force; in this way, the stress state at the blade root is closer to the actual working condition during biaxial dynamic loading.

[0137] Furthermore, when extracting the deformation boundary line of the inner wall of the casing in S5, for the reference image before deformation and each frame image after deformation, the displacement field of each sub-region is calculated using a local sub-region matching method. The specific matching method is as follows:

[0138] The image is divided into multiple overlapping sub-regions, each of which is a square pixel block. The gray-level distribution of one sub-region in the reference image is selected as a template. The sub-region position with the highest correlation to the gray-level distribution of the template is searched in the deformed image. The displacement vector of the center of the sub-region is obtained by the offset of the position relative to the original position of the template.

[0139] Interpolate the displacement vectors of all sub-region centers to obtain a displacement field cloud map of the entire inner wall surface of the casing. Extract continuous pixel bands whose displacement gradient exceeds the gradient threshold from this displacement field cloud map as deformation boundary lines.

[0140] In this embodiment, during the test preparation stage, a random speckle pattern is pre-coated onto the inner wall surface of the casing. Specifically, the inner wall of the casing is first sandblasted to remove the oxide layer and increase surface roughness. Then, a black high-temperature resistant coating, primarily composed of carbon black and silicone resin, and a white high-temperature resistant coating, primarily composed of titanium dioxide and silicone resin, are selected. The two coatings are respectively loaded into the canisters of two air spray guns, and the air pressure of the spray guns is adjusted to a fixed value, for example, 0.5 MPa. The distance between the spray gun nozzle and the inner wall surface of the casing is controlled according to a fixed value, for example, 300 mm. The operator first uses the white coating to evenly spray a layer of... Apply a layer of white primer. After the primer dries, apply black paint using a spray gun in a dotted spray pattern, creating randomly distributed spots on the white primer. During spraying, move the spray gun randomly to avoid forming periodic patterns. Adjust the nozzle's needle valve opening so that the average diameter of the black spots occupies three to five pixels in the image – this can be calibrated by test spraying and taking sample images: take a speckle image at a standard distance, measure the pixel diameter of each spot, and if most spots are between three and five pixels in diameter, record the current spray gun pressure and distance parameters for use in the actual spraying.

[0141] When extracting the deformation boundary line of the inner wall of the casing in S5, for the reference image before deformation (i.e., the last frame before fracture excitation) and each frame after deformation, the displacement field of each sub-region is calculated using a local sub-region matching method. The specific matching method is as follows: the image is divided into multiple overlapping square sub-regions, the size of which is, for example, 31 pixels by 31 pixels, and the step size between the centers of adjacent sub-regions is 5 pixels. In the reference image, the gray distribution of one of the sub-regions is selected as the template, i.e., a 31 by 31 gray matrix. In the deformed image, with the position of the center of the sub-region in the reference image as the center, within a preset search area, such as 41 pixels by 41 pixels, a sub-region of the same size is moved pixel by pixel, and the correlation between the template and the gray distribution of the current sub-region is calculated. The correlation is measured by the zero-mean normalized cross-correlation coefficient. The position of the sub-region with the largest correlation coefficient is found. The offset of this position relative to the original position of the template is the displacement vector of the center of the sub-region, which includes row displacement and column displacement.

[0142] Repeat the above operation for all sub-regions to obtain the displacement vector of the center point of each sub-region. Then, use bicubic spline interpolation between the center points of each sub-region to obtain the displacement field cloud map of the entire inner wall surface of the casing—that is, the displacement vector at each pixel position; extract the continuous pixel bands with displacement gradients exceeding the gradient threshold from the displacement field cloud map as deformation boundary lines; the displacement gradient is calculated as follows: for each pixel, calculate the partial derivatives of its displacement components in the row and column directions, combine them into a gradient tensor, and take the maximum principal strain of the tensor as the gradient magnitude of that point; set a gradient threshold, which is determined according to the yield strain of the inner wall material of the casing: for example, obtain the yield strain value of the material through a material tensile test, and multiply it by a safety factor, such as 0.8, as the threshold; mark the pixels with gradient magnitudes greater than the threshold as high gradient points; then use an edge tracking algorithm to connect these high gradient points into a continuous pixel band—starting from a high gradient point, find adjacent high gradient points along the gradient direction until it is impossible to continue; the extracted continuous pixel bands are the deformation boundary lines, which usually appear as a closed loop or multiple arc-shaped strips around the impact center.

[0143] According to a second aspect of the present invention, the present invention claims protection for a high-speed camera monitoring system for electromagnetic dynamic dual-axis synchronous loading, comprising:

[0144] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0145] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0146] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading, characterized in that, include: S1: The turbine engine rotor component to be monitored is installed on the rotor support structure of the electromagnetic dynamic biaxial loading test bench. Radial electromagnetic loaders are arranged in the radial direction and axial electromagnetic loaders are arranged in the axial direction of the test bench. Multiple high-speed camera groups with optical spindles pointing to the preset debris scattering space area are arranged around the outer casing. S2: By sending loading commands to the radial electromagnetic loader and the axial electromagnetic loader respectively through an independent timing controller, the two electromagnetic loaders synchronously apply dynamically changing electromagnetic forces to generate a dual-axis synchronous preload on the rotor component that simulates the operating conditions of an engine. S3: Activate the fracture excitation device installed at the predetermined fracture site of the rotor component, so that the predetermined fracture site will fracture instantaneously under the combined action of biaxial preload, releasing multiple fragments. S4: The timing controller sends synchronous trigger pulses to all high-speed camera groups to acquire images from the moment the fragment breaks until the fragment hits the inner wall of the casing and stops moving, recording the instantaneous image of the fragment in space and the deformation stripe image of the inner wall surface of the casing. S5: Extract the geometric centroid coordinates of the fragments and the deformation boundary line of the inner wall of the casing frame by frame from the image sequence, and use the pre-calibrated spatial coordinate transformation matrix to obtain the flight trajectory coordinate sequence of the fragments in three-dimensional space and the boundary of the impact indentation area of ​​the inner wall of the casing. S6: Calculate the incident velocity vector and the exit velocity vector after penetration at the impact point of the fragment on the inner wall of the casing based on the flight trajectory coordinate sequence. Determine whether the fragment has penetrated the casing based on whether the exit velocity vector is zero, and record the direction of the remaining velocity vector when penetration occurs.

2. The method according to claim 1, characterized in that, The radial electromagnetic loader described in S1 includes a plurality of radial electromagnets evenly distributed along the circumference of the rotor component. The core axis of each radial electromagnet coincides with the radial direction of the rotor component, and a non-contact air gap is maintained between the core end face of each radial electromagnet and the outer edge surface of the rotor component. The axial electromagnetic loader includes at least two coaxially arranged annular electromagnets, which are respectively disposed on the front end face and the rear end face of the rotor component. The inner diameter of each annular electromagnet is larger than the diameter of the rotor shaft of the rotor component. The end face of the annular electromagnet and the corresponding end face of the rotor component also maintain a non-contact air gap. The inner wall surface of the outer casing is coated with a diffuse reflection coating, which is formed by spraying a mixture of magnesium oxide powder and epoxy resin in a certain proportion, and is used to enhance the texture contrast of the inner wall of the casing in high-speed camera images.

3. The method according to claim 1, characterized in that, The dynamically changing electromagnetic force described in S2 is a time-varying force in the form of a sine wave. The electromagnetic forces applied by the radial electromagnetic loader and the axial electromagnetic loader have the same oscillation frequency, and the phase difference between the electromagnetic forces in the two directions is preset according to the force phase relationship of the rotor components in actual engine operation. The timing controller contains two independent waveform generation channels. Each channel outputs a programmable voltage waveform to the drive coil of the corresponding electromagnetic loader. The two channels are locked in phase by a clock synchronization signal, so that the radial electromagnetic force and the axial electromagnetic force can change in tandem according to a preset phase difference. During the application of the dual-axis synchronous preload, the magnitude of the electromagnetic force gradually increases from zero to the target amplitude. The rate of increase is determined according to the material strength and fracture toughness of the rotor components to avoid unexpected premature fracture before the fracture excitation device is activated.

4. The method according to claim 1, characterized in that, S4 specifically includes the following sub-steps: S41: Generates rising edge synchronous trigger pulses inside the timing controller and transmits them in parallel to the external trigger interface of each high-speed camera group via coaxial cable; S42: Each high-speed camera group is equipped with a high-precision crystal oscillator. When the external trigger interface receives the rising edge of the synchronous trigger pulse, the crystal oscillator starts counting and starts the first frame exposure of the camera group after reaching the preset delay count. The delay count is independently calibrated according to the optical path distance between each camera group and the fracture excitation position. S43: Each high-speed camera group continuously performs exposure and readout operations for subsequent frames according to a pre-set fixed frame rate. The interval between the exposure start times of two adjacent frames is obtained by frequency division of the crystal oscillator. The fixed frame rate is determined based on the expected maximum speed of the debris flying in the internal space of the housing and the maximum allowable motion blur displacement of the debris in each frame image. S44: A narrow-band filter is installed in front of the photosensitive surface of the image sensor of each high-speed camera group. The center wavelength is matched with the emission wavelength of the stroboscopic illumination source arranged around the test bench. During the exposure of each frame image, the timing controller sends a pulse drive current synchronized with the exposure window of the camera group to the stroboscopic illumination source. The edge contour of the fragment and the deformation characteristics of the inner wall surface of the casing are recorded. S45: Each high-speed camera group continuously acquires images until it detects that the change in the position of a fragment in multiple consecutive frames is less than the spatial distance corresponding to one pixel. It then automatically stops acquiring images and stores the acquired complete image sequence in the non-volatile memory inside the high-speed camera group.

5. The method according to claim 1, characterized in that, The pre-calibrated spatial coordinate transformation matrix described in S5 is obtained in the following way: A calibration target is placed in the debris scattering area, and multiple high-speed camera groups simultaneously capture images of the calibration target to extract the two-dimensional pixel coordinates of the spherical marker points. For each camera group, establish the projection relationship between the two-dimensional pixel coordinates extracted from the image of the group and the three-dimensional coordinates of the corresponding marker points on the calibration target, and solve for the projection matrix corresponding to the group. By combining the projection matrices of all camera groups into a set of three-dimensional spatial coordinate transformation equations, the transformation matrix from the two-dimensional pixel coordinates of any camera group to the unified three-dimensional spatial coordinates can be obtained.

6. The method according to claim 1, characterized in that, The incident velocity vector described in S6 is calculated in the following way: The three-dimensional coordinates of the frame before the fragment contacts the inner wall of the casing and the three-dimensional coordinates of the frame before the contact are selected from the fragment flight trajectory coordinate sequence. The difference vector is divided by the time interval between the two frames to obtain the incident velocity vector. The ejection velocity vector is calculated in the following way: Select the three-dimensional coordinates of the first frame after the fragment contacts the inner wall of the casing and the three-dimensional coordinates of the second frame after contact from the fragment flight trajectory coordinate sequence. Divide the difference vector by the time interval between the two frames to obtain the ejection velocity vector. If the three-dimensional coordinates of the first frame after contact coincide with the three-dimensional coordinates of the second frame after contact, the ejection velocity vector is determined to be zero, indicating that the fragment failed to penetrate the casing. If the launch velocity vector is non-zero, the direction angle of that vector is recorded as the direction of the remaining velocity vector. This direction angle includes the launch elevation angle relative to the radial surface of the casing and the launch azimuth angle relative to the circumferential direction of the casing.

7. The method according to claim 1, characterized in that, S5 includes: The method for identifying the pixel regions occupied by debris in images captured by a high-speed camera group is as follows: Traverse all pixels in the image, compare the gray value of each pixel with a dynamic threshold, and classify pixels with gray values ​​greater than the dynamic threshold as bright areas. The connected region with the largest area in the bright area is determined as the fragment region. Calculate the average coordinate of all pixels in the fragment region as the two-dimensional pixel coordinate of the fragment in the image, and record the pixel area of ​​the fragment region. If the pixel area changes by more than twice between two consecutive frames, it is determined that the fragment broke or collided with the inner wall of the casing at that frame. For the same sampling time, multiple two-dimensional pixel coordinates are obtained. Using a pre-calibrated projection matrix, these coordinates are converted into spatial rays emanating from the viewpoint of the camera group. The coordinates of the spatial point with the minimum sum of squared distances between all rays are then calculated. The calculation method is as follows: The ray equation corresponding to each camera group is written in the form of the distance from a point to a line. An objective function is established, and the spatial coordinates that minimize the objective function are found through numerical iteration. The spatial coordinates of the spatial point are used as the three-dimensional spatial coordinates of the fragment at that sampling time. The three-dimensional spatial coordinates obtained from multiple consecutive sampling times are arranged to form a discrete spatial position sequence of fragments. For each intermediate point in the sequence, the displacement vector between it and the previous point and the average velocity vector corresponding to the displacement vector are calculated. Then, the average velocity vector corresponding to the displacement vector between it and the next point is calculated. The average value of these two average velocity vectors is taken as the instantaneous velocity vector estimate at that point. The end points of all instantaneous velocity vectors are connected to form a velocity vector chain. For the deformation boundary line of the inner wall of the casing, extract the intersection points of the pre-made grid marking lines on the surface of the inner wall of the casing. The method for extracting the intersection points is as follows: Search for the continuous pixel line with the largest gray-level gradient along the row and column directions in the image, and determine the intersection point as the intersection point to obtain the two-dimensional pixel coordinates of each intersection point. The two-dimensional coordinates of the intersection point are converted into three-dimensional spatial coordinates using a projection matrix. After the consistency check is performed on the three-dimensional spatial coordinates obtained by all camera groups, the average value is taken as the final three-dimensional spatial coordinates of the intersection point. Compare the three-dimensional spatial coordinates of the intersection of the same grid marking line before deformation with the three-dimensional spatial coordinates after deformation, calculate the displacement vector of each intersection, and the area enclosed by the intersections whose displacement vector magnitude exceeds the preset threshold is the edge of the impact depression area. The discrete spatial position sequence of the fragments, the velocity vector chain, and the closed polygon of the impact depression region boundary are stored together in the data recording unit of the test bench as input data for subsequent judgment of fragment penetration behavior.

8. The method according to claim 1, characterized in that, The rotor support structure described in S1 includes a detachable rotor simulation disk. The central shaft is supported on the base of the test bench by a magnetic levitation bearing. The stator part of the magnetic levitation bearing is fixed inside the base, and the rotor part is sleeved on the central shaft. During loading, the magnetic levitation bearing actively adjusts the radial position of the central shaft. The outer edge of the rotor simulation disk is provided with multiple slots for mounting the rotor components to be tested. Each slot has a strain gauge at the bottom for measuring the preload. The opening direction of the slot is consistent with the radial direction of the rotor simulation disk. When installing the rotor component to be tested, insert the root of the rotor component into the slot and lock it with a wedge block. The locking torque of the wedge block is adjusted to a preset force range according to the strain gauge reading. This force range corresponds to the equivalent value of the centrifugal force borne by the rotor component in the actual engine.

9. The method according to claim 1, characterized in that, When extracting the deformation boundary line of the inner wall of the casing in S5, for the reference image before deformation and each frame image after deformation, the displacement field of each sub-region is calculated using the local sub-region matching method. The specific matching method is as follows: The image is divided into multiple overlapping sub-regions, each of which is a square pixel block. The gray-level distribution of one sub-region in the reference image is selected as a template. The sub-region position with the highest correlation to the gray-level distribution of the template is searched in the deformed image. The displacement vector of the center of the sub-region is obtained by the offset of the position relative to the original position of the template. Interpolate the displacement vectors of all sub-region centers to obtain a displacement field cloud map of the entire inner wall surface of the casing. Extract continuous pixel bands whose displacement gradient exceeds the gradient threshold from this displacement field cloud map as deformation boundary lines.

10. A high-speed camera monitoring system for electromagnetic dynamic dual-axis synchronous loading, characterized in that, include: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement a high-speed camera monitoring method for electromagnetic dynamic dual-axis synchronous loading according to any one of claims 1 to 9.

Images