Magnetic control navigation system based on permanent magnet array

Abstract

The invention belongs to the field of medical instruments, and particularly relates to a magnetic control navigation system based on a permanent magnet array. The system comprises two movable magnetic arrays, each magnetic array is provided with a first adjusting mechanism, a second adjusting mechanism and a third adjusting mechanism, full-angle adjustment of a magnetic ball is achieved, and the magnetic field adjusting capacity of the magnetic control navigation system is improved; the position of the magnetic ball is adjusted through the third adjusting mechanism, so that inconvenience caused by moving the mechanical arm is avoided, the flexibility of magnetic field adjustment of the magnetic control navigation system is further improved, and the occupied volume of the magnetic control navigation system is also reduced; an electromagnet is not used, so that a cooling system does not need to be arranged; the optimal distribution mode of the magnetic array can be found by using the pre-trained neural network model, the magnetic field intensity and the isotropy are improved, and through step-by-step iterative optimization and neural network assistance, the calculation complexity is reduced, and the system response speed is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of medical devices, and particularly relates to a magnetic navigation system based on a permanent magnet array. BACKGROUND

[0002] At present, the most common magnetic navigation system for neurointervention adopts electromagnets or permanent magnets to generate an excitation magnetic field, including an electromagnet array magnetic navigation system and a magnetic navigation system based on a mechanical arm driving permanent magnet.

[0003] The electromagnet array magnetic navigation system can adjust the strength of the magnetic field by changing the size of the current and the number of turns of the external coil, and can realize the adjustable function of the direction of the magnetic field by deploying multiple electromagnets in different directions. However, the magnetic energy density of the electromagnet is low, and in order to generate a strong enough magnetic field, the device will occupy a large volume, which is not suitable for conventional operating rooms. In addition, when a large current passes through the coil, a large amount of heat will be generated, in order to ensure that the system works at a proper temperature, a cooling system is usually equipped to dissipate heat, and the system cannot work for a long time.

[0004] For the magnetic navigation system based on the mechanical arm driving permanent magnet, the dynamic control of the magnetic field is mainly realized by the macro-motion of the mechanical arm and the multi-degree-of-freedom rotation of the end permanent magnet. However, the inherent limitations of such systems significantly affect their performance: first, the large structural weight and motion inertia of the mechanical arm result in slow dynamic response, making it difficult to achieve rapid and flexible magnetic field changes; second, in order to obtain sufficient magnetic field strength in the working space, the mechanical arm end often needs to carry a large size permanent magnet. For such a large single permanent magnet, when the distance between the geometric center of the permanent magnet and the working target point is less than twice the effective radius, the magnetic field prediction based on the magnetic dipole model will introduce significant errors, which directly weakens the accuracy of magnetic field control. SUMMARY

[0005] In order to solve at least one of the technical problems mentioned in the background, the purpose of the present application is to provide a magnetic navigation system based on a permanent magnet array, which can solve the problem of large volume of existing magnetic navigation systems, and does not need to be equipped with a cooling system, thereby improving the flexibility and control accuracy of the magnetic field change.

[0006] The technical scheme provided by the present application is as follows: A magnetic navigation system based on a permanent magnet array, comprising: two mirror image arranged liftable support structures; Two movable magnetic arrays are respectively mounted on top of two support mechanisms; each magnetic array includes a base and multiple magnetic units distributed on its surface, with the magnetic units in the two magnetic arrays arranged opposite each other along a vertical plane; each magnetic unit includes a permanent magnet-type magnetic ball, a first adjustment mechanism, a second adjustment mechanism, and a third adjustment mechanism; the first adjustment mechanism is used to drive the magnetic ball to rotate in a direction perpendicular to the magnetic axis; the second adjustment mechanism is used to drive the magnetic ball to rotate in a rotation direction perpendicular to the first adjustment mechanism; the third adjustment mechanism is used to drive the magnetic ball to pitch along a vertical plane to adjust the position of the magnetic ball. The magnetic field control module is electrically connected to the magnetic array. The module includes a spatial optimization unit and an execution unit. The spatial optimization unit runs a pre-trained neural network model, which predicts the performance indicators of the magnetic field formed by each magnetic sphere in the magnetic array under different pitch states, including magnetic field strength and isotropy. The spatial optimization unit first iteratively optimizes the optimal spatial distribution of the magnetic array using the neural network model. Then, under the optimal spatial distribution, it uses a control algorithm to inversely solve for the rotation angle of each magnetic sphere that enables the formation of the desired target magnetic field between the two magnetic arrays. The execution unit first issues commands to each third adjustment mechanism to adjust the pitch angle of each magnetic unit, so that the magnetic array is in the optimal spatial distribution state. Then, it issues commands to the first and second adjustment mechanisms to adjust the rotation angle of each magnetic unit to obtain the desired magnetic field distribution.

[0007] As a further improvement of the present invention, the first adjustment mechanism includes a first rotating shaft, an outer support, and a first driver; the magnetic ball is movably connected to the outer support via a first rotating shaft perpendicular to the magnetic axis of the magnetic ball; the first driver is drivenly connected to the first rotating shaft. The second adjustment mechanism includes a second rotating shaft, a main support, and a second driver; the outer support is rotatably connected to the main support via a second rotating shaft perpendicular to the first rotating shaft; the second driver is drive-connected to the second rotating shaft. The third adjustment mechanism includes a pitch rotation shaft, a pitch driver, and an angle lock; the pitch rotation shaft is fixed to the main support in the horizontal direction and is hinged to the outer support; the pitch driver is fixed to the main support and is connected to the outer support in a transmission manner; the angle lock is located at the end of the pitch rotation shaft and locks the angle of the outer support by increasing friction or mechanical latching.

[0008] As a further improvement of the present invention, the outer support includes two parallel connecting rods and a bracket connecting the ends of the two connecting rods on the same side; the connecting rods are perpendicular to the first rotation axis and parallel to the second rotation axis; the base of the first driver is fixedly connected to one of the connecting rods, so that the first driver rotates in a direction perpendicular to the two connecting rods; the magnetic ball rotates by the rotation of the first driver which is connected to the first rotation axis; the bracket is rotatably connected to the main support through a ball bearing.

[0009] As a further improvement of the present invention, the first driver and the first rotating shaft are connected by a gear assembly; the gear assembly includes: a driving wheel, a transmission wheel and a driven wheel; the transmission wheel includes two layers of gears, the first layer of gears meshing with the driving wheel and the second layer of gears meshing with the driven wheel; the driving wheel is fixedly connected to the first driver; the driven wheel is fixedly connected to the first rotating shaft.

[0010] As a further improvement of the present invention, the second driver and the second rotating shaft are connected by a support assembly; the support assembly includes: a rotating support vertically connected between two connecting rods, and a drive support fixedly connected between the second driver and the bracket; the rotating support is fixedly connected to the base of the second driver, so that the second driver rotates in a direction perpendicular to the rotating support; the drive support drives the outer support to rotate through the rotation of the second driver.

[0011] As a further improvement of the present invention, the magnetic array includes nine magnetic units arranged in a 3×3 configuration; for each column of magnetic units, three magnetic balls are located on the same circumscribed circle, and the angles between the line connecting the center of the magnetic balls on both sides to the center of the circumscribed circle and the horizontal line are 57° and -47°, respectively; the angle between the line connecting the center of the magnetic ball in the middle to the center of the circumscribed circle and the horizontal line is 5°.

[0012] As a further improvement to the present invention, the design method of the neural network model includes: S1: Establish a mapping model, which pre-defines the mapping relationship between the magnet array configuration and the magnetic field characteristics; S2: Design the objective function with magnetic field strength and isotropy as optimization objectives; use numerical optimization algorithms and mapping models to iteratively optimize the magnetic space parameters of the magnetic sphere to obtain the objective function values ​​of each round of optimization; combine the input magnetic space parameters and output objective function values ​​in each round of optimization into a data pair; S3: Treat each data pair as a sample data to obtain a dataset containing a large number of sample data; divide the dataset into training set, test set, and validation set, train the pre-built neural network model until the loss function converges, and obtain the trained neural network model.

[0013] As a further improvement of the present invention, step S2, which involves iteratively optimizing the magnetic spatial parameters of the magnetic sphere using a numerical optimization algorithm and a mapping model to obtain the objective function values ​​for each round of optimization, includes: S21: Gradually adjust the rotation angle of one of the magnetic units and obtain the corresponding objective function value. Search for the rotation angle corresponding to the minimum objective function value and use it as the target angle of the current magnetic unit. S22: Fix the magnetic unit with the known target angle to the corresponding state, and use the method of S21 to determine the target angle of each of the remaining magnetic units in sequence; S23: The objective function value with all magnetic units at the target angle is taken as the final optimization result.

[0014] As a further improvement to the present invention, the numerical optimization algorithm is the Nelder-Mead polyhedron search algorithm.

[0015] As a further improvement of the present invention, the neural network model is a multilayer perceptron neural network model.

[0016] The technical solution provided by this invention has the following beneficial effects: This invention provides a magnetically controlled navigation system based on a permanent magnet array. The magnetic array is equipped with a first adjustment mechanism and a second adjustment mechanism to achieve full-angle adjustment of the magnetic ball, thereby improving the magnetic field adjustment capability of the magnetically controlled navigation system. The position of the magnetic ball is adjusted by a third adjustment mechanism, avoiding the inconvenience caused by moving the robotic arm, further improving the flexibility of magnetic field adjustment of the magnetically controlled navigation system, and also reducing the volume occupied by the magnetically controlled navigation system. Since no electromagnets are used, there is no need to equip it with a cooling system. The optimal distribution mode of the magnetic array can be found using a pre-trained neural network model to improve the magnetic field strength and isotropy. Through step-by-step iterative optimization and neural network assistance, the computational complexity is reduced and the system response speed is improved. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall structure of a magnetically controlled navigation system based on a permanent magnet array provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the magnetic array structure of a magnetically controlled navigation system based on a permanent magnet array, as provided in Embodiment 1 of the present invention. Figure 3 This is a structural diagram of a magnetic field control module provided in Embodiment 1 of the present invention; Figure 4 This is a top view of a magnetic unit provided in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of a magnetic array in its optimal spatial distribution state as provided in Embodiment 1 of the present invention; Figure 6 This is a flowchart of a neural network model design method provided in Embodiment 2 of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] Example 1 This embodiment provides a magnetically controlled navigation system based on a permanent magnet array, such as... Figure 1 As shown, Figure 1 This is a schematic diagram of the overall structure of a magnetically controlled navigation system based on a permanent magnet array, as provided in Embodiment 1 of the present invention. The system includes: Two mirror-mounted, height-adjustable support structures 101 are positioned on either side of the target object, symmetrically distributed along the target object, with adjustable spacing between the two support structures 101. A lifting mechanism 1011 is located at the neck of each support structure, with a lifting stroke covering a height range of 0.5 to 3 meters. Synchronous lifting is achieved by a servo motor driving a lead screw, ensuring alignment of the magnetic array working plane with the navigation target area, and enabling precise control of the magnetic field strength and direction of the target area.

[0020] Two movable magnetic arrays 102 are respectively positioned on top of two supporting structures 101. In practical applications, such as in neurointerventional surgery, it is necessary to generate a high-intensity and highly uniform magnetic field in the target vascular region to precisely control the movement of the guidewire. The two symmetrically distributed magnetic arrays 102 work together to form a superimposed magnetic field. The strength of the magnetic field is significantly increased after superposition, and the symmetrical layout can also counteract the non-uniformity of the magnetic field on one side, reducing the "weak field area at the edge".

[0021] like Figure 2 As shown, Figure 2 This is a schematic diagram of the magnetic array structure of a magnetically controlled navigation system based on a permanent magnet array, as provided in Embodiment 1. Each magnetic array includes a base 1021 and multiple magnetic units 1022 arranged in an array on the surface of the base. Figure 2 The diagram shows nine magnetic units 1022 arranged in a 3×3 configuration. Each magnetic unit 1022 includes a permanent magnet-type magnetic ball, a first adjustment mechanism, a second adjustment mechanism, and a third adjustment mechanism. The first adjustment mechanism drives the magnetic ball to rotate in a direction perpendicular to the magnetic axis, and the second adjustment mechanism drives the magnetic ball to rotate in a direction perpendicular to the rotation of the first adjustment mechanism. The magnetic axis is the line connecting the N and S poles of the magnetic ball. The first and second adjustment mechanisms provide two degrees of freedom for the rotational adjustment of the magnetic ball, thereby changing the magnetic field distribution by adjusting the magnetic pole direction of the magnetic ball at all angles, expanding the magnetic field adjustment capability of the system. The third adjustment mechanism indirectly adjusts the position of the magnetic ball by driving the magnetic unit 1022 to pitch along a vertical plane. Each magnetic ball can be independently controlled, forming a three-dimensional magnetic field coverage network. In practical applications, the magnetic field gradient distribution better conforms to the three-dimensional anatomical structure of the target object, such as three-dimensionally covering the complex three-dimensional path of blood vessels, overcoming the two-dimensional limitations of traditional planar arrays.

[0022] Magnetic field control module ( Figure 1(Not shown), it is electrically connected to the magnetic array 102. "Electrical connection" can be via electrical components such as terminals or PCB connectors, or it can be via signal transmission. In this magnetic navigation system, the position and orientation of the magnetic poles of multiple magnetic spheres directly affect the strength, uniformity, and controllability of the magnetic field. Therefore, it is necessary to find the optimal arrangement of the magnets so that the magnetic field can accurately control the guided wire, magnetic particles, and other controlled objects, without causing the controlled objects to lose control due to sudden changes in the magnetic field.

[0023] like Figure 3 As shown, Figure 3 This is a structural diagram of a magnetic field control module provided in Embodiment 1 of the present invention. The magnetic field control module 201 includes a spatial optimization unit 2011 and an execution unit 2012. The spatial optimization unit 2011 runs a pre-trained neural network model, which can predict the performance indicators of the magnetic field formed by each magnetic sphere in the magnetic array under different pitch states. The performance indicators of the magnetic field include magnetic field strength and isotropy, where isotropy means that the magnetic field has the same physical properties (such as strength and gradient) in all directions in space. In this embodiment of the invention, optimizing the magnetic field isotropy means making the magnetic force on the guide wire uniform in any direction, avoiding control deviations caused by direction dependence.

[0024] The spatial optimization unit 2011 first iteratively optimizes the optimal spatial distribution of the magnetic array using a neural network model. Then, under the optimal spatial distribution, it uses a control algorithm to inversely calculate the rotation angle of each magnetic sphere that enables the formation of the desired target magnetic field between the two magnetic arrays. This rotation angle includes rotation angles in two directions. The execution unit 2012 first issues commands to each of the third adjustment mechanisms to adjust the pitch angle of each magnetic unit so that the magnetic array is in the optimal spatial distribution. Then, based on the rotation angle calculated by the spatial optimization unit 2011, it issues commands to the first and second adjustment mechanisms to adjust the rotation angle of each magnetic unit to obtain the desired magnetic field distribution.

[0025] Operators can input commands such as the magnitude, direction, and position of the magnetic field using an XBOX controller. These commands are wirelessly transmitted via Bluetooth to the interactive software interface in the Ubuntu system and displayed. After the operator confirms the information, the rotation angle data of each magnetic ball is calculated by the control algorithm in the Ubuntu system by setting the target value. The rotation angle data is then transmitted to the STM32 controller via CAN (Controller Area Network) communication. After receiving the data, the STM32 controller distributes it to different PCA9685 servo control boards. Finally, the servo receives the corresponding PWM signal and performs rotation braking, generating the target magnetic field and driving the magnetic soft guide wire to deflect.

[0026] likeFigure 4 As shown, Figure 4 This is a top view of a magnetic unit provided in Embodiment 1 of this application.

[0027] The first adjustment mechanism includes a first rotating shaft 301, an outer support (including two parallel connecting rods 3021 and a bracket 3022 connecting the two connecting rods to the same end), and a first driver 303. The magnetic ball is movably connected to the connecting rod 3021 through a first rotating shaft 301 perpendicular to the magnetic axis direction of the magnetic ball. The first rotating shaft 301 is perpendicular to the connecting rod 3021.

[0028] The first rotating shaft 301 can be tightly connected to the two connecting rods 3021 through ceramic ball bearings. The advantages of using rolling bearings made of ceramic materials (such as silicon nitride) include: non-magnetic, which does not affect the magnetic field distribution of permanent magnets; corrosion resistance, which is suitable for medical sterilization environments; high hardness, which reduces wear and extends service life.

[0029] The first actuator 303 can be a 150kg servo motor, with its base fixedly connected to one of the connecting rods 3021, allowing the first actuator 303 to rotate in a direction perpendicular to both connecting rods 3021. The first actuator 303 is driven by the first rotating shaft 301, specifically through a gear assembly. The gear assembly includes a drive wheel 3041, a transmission wheel 3042, and a driven wheel 3043. The transmission wheel 3042 includes two layers of gears; the first layer meshes with the drive wheel 3041, and the second layer meshes with the driven wheel 3043. The first actuator 303 is fixedly connected to the drive wheel 3041, and the driven wheel 3043 is fixedly connected to the first rotating shaft 301. The secondary transmission of the transmission wheel 3042 drives the first rotating shaft 301. The magnetic ball is axially positioned at both ends through the shoulders of the first rotating shaft 301 and double nuts. The magnetic ball is circumferentially positioned by a square hole fit between it and the first rotating shaft 301, allowing the magnetic ball to rotate together with the first rotating shaft 301.

[0030] The second adjustment mechanism includes a second rotating shaft 305, a main support 306, and a second actuator 307. The bracket 3022 is rotatably connected to the main support 306 via a second rotating shaft 305 perpendicular to the first rotating shaft 301. Specifically, a deep groove ball bearing can be used to achieve a tight connection with the main support 306, providing stable support and preventing wear caused by direct contact between mechanical parts. The second actuator 307 is connected to the second rotating shaft 305 via a support assembly. The support assembly includes a rotating bracket 3081 vertically connected between two connecting rods 3021, and a drive bracket 3082 fixedly connected between the second actuator 307 and the bracket 3022. The second actuator 307 can be a 70kg servo motor. The base of the second actuator 307 is fixedly connected to the rotating bracket 3081, allowing the second actuator 307 to rotate in a direction perpendicular to the rotating bracket 3081, and then driving the entire outer support to rotate via the drive bracket 3082. The first and second adjustment mechanisms cooperate to achieve dual-degree-of-freedom adjustment of the magnetic ball, improving the system's magnetic field adjustment capability.

[0031] The third adjustment mechanism includes a pitch rotation shaft, a pitch actuator, and an angle lock (not shown in the figure). The pitch rotation shaft is fixed horizontally to the main support 306 and is hinged to the outer support; the pitch actuator is fixed to the main support 306 and is connected to the outer support via a transmission; the angle lock is located at the end of the pitch rotation shaft and locks the angle of the outer support by increasing friction or mechanically snapping it shut.

[0032] Typically, assume that the magnetic array 102 comprises nine magnetic elements 1022 arranged in a 3×3 configuration. Figure 5 This diagram illustrates a magnetic array in its optimal spatial distribution state according to Embodiment 1 of this application. As shown in the diagram, for each column of magnetic units 1022, three magnetic spheres are located on the same circumscribed circle. For ease of description, the three magnetic units are referred to as A, B, and C from top to bottom. In magnetic unit A, the angle between the line connecting the center of the magnetic sphere to the center of the circumscribed circle and the horizontal line is 57°, and the angle between the central axis of magnetic unit A and the horizontal line is 60°. In magnetic unit B, the angle between the line connecting the center of the magnetic sphere to the center of the circumscribed circle and the horizontal line is 5°, and the angle between the central axis of magnetic unit B and the horizontal line is 60°. In magnetic unit C, the angle between the line connecting the center of the magnetic sphere to the center of the circumscribed circle and the horizontal line is -47°, and the angle between the central axis of magnetic unit C and the horizontal line is 15°.

[0033] The application scenarios of a magnetically controlled navigation system based on a permanent magnet array provided in this invention include, but are not limited to: (1) Guiding and positioning: In vascular interventional surgery, magnetic fields can be used to guide catheters or other medical devices to move precisely to the target location. By interacting with small magnets on the device through an external magnetic field, the device can be manipulated more precisely.

[0034] (2) Targeted drug delivery: By using the precise gradient magnetic field generated by the magnetic navigation system, drugs can be delivered directly to specific areas of the body more effectively, reducing potential damage to surrounding healthy tissues.

[0035] (3) Thrombus removal: By generating a relatively uniform and strong magnetic field, the magnetic catheter and guidewire are controlled to reach the site of cerebral vascular occlusion, and the guidewire is driven to remove or break up the thrombus in the blood vessel.

[0036] (4) Cardiac defibrillation: In the treatment of certain heart diseases, the magnetic field generated by the magnetic navigation system can be used to precisely guide the treatment device, such as in heart valve surgery, to guide the magnetically controlled catheter and guidewire to the lesion quickly and accurately.

[0037] In summary, this invention provides a magnetically controlled navigation system based on a permanent magnet array. The magnetic array is equipped with a first adjustment mechanism and a second adjustment mechanism to achieve full-angle adjustment of the magnetic ball, thereby improving the magnetic field adjustment capability of the magnetically controlled navigation system. The third adjustment mechanism adjusts the position of the magnetic ball, avoiding the inconvenience caused by moving the robotic arm, further improving the flexibility of the magnetic field adjustment of the magnetically controlled navigation system, and also reducing the volume occupied by the magnetically controlled navigation system. Since no electromagnet is used, there is no need to equip it with a cooling system.

[0038] Furthermore, for such large-sized single permanent magnets, when the distance between their geometric center and the target point is less than twice their effective radius (i.e., entering the near-field region), magnetic field prediction based on the magnetic dipole model introduces significant errors. However, the magnetic navigation system based on a permanent magnet array provided in this invention cleverly circumvents this limitation through structural innovation: in a typical case, magnetic spheres with a radius of approximately 30mm are used, resulting in a larger effective working area compared to traditional large magnets. This characteristic is particularly important in delicate surgical scenarios such as head vascular intervention. When the target point is close to a magnetic sphere, other magnetic units remain in their respective far-field regions, and their magnetic field distribution can still be accurately predicted using the magnetic dipole model, thereby reducing prediction errors. Through array-based combination and spatially optimized layout, when several magnetic spheres are in the near-field range of the target point, the remaining magnetic spheres in the far-field state can still provide high-precision magnetic field components, reducing global prediction deviation through mutual error cancellation. This distributed magnetic source design provides a foundation for accurate magnetic field modeling and flexible control, thus significantly improving the control accuracy and dynamic performance of the navigation system.

[0039] Example 2 like Figure 6 As shown, Figure 6 A flowchart illustrating a design method for a neural network model provided in Embodiment 2 of this application. Based on Embodiment 1, the specific design method for the aforementioned neural network model is as follows: S1: Establish a mapping model, which pre-determines the mapping relationship between the magnet array configuration and the magnetic field characteristics.

[0040] In a magnetic array, the pitch of the magnetic elements determines the spatial position and rotation axis direction of the internal magnetic spheres. Different configurations of magnetic arrays exhibit different magnetic field characteristics. A model describing the mapping relationship between the magnetic array configuration and magnetic field characteristics is established beforehand as the theoretical basis for subsequent optimization.

[0041] S2: An objective function is designed with magnetic field strength and isotropy as optimization targets. Numerical optimization algorithms and mapping models are used to iteratively optimize the magnetic spatial parameters of the magnetic sphere, obtaining the objective function values ​​for each optimization round. The input magnetic spatial parameters and output objective function values ​​from each optimization round are combined into a data pair. The magnetic spatial parameters of the magnetic sphere are its spatial position and rotation axis direction, which are determined by the pitch of the magnetic unit.

[0042] The above is the initial optimization based on the physical model. However, calculations based on the physical model are very time-consuming. Therefore, it is necessary to collect data from this stage of optimization to train a neural network model. This allows the trained neural network model to quickly predict the objective function value of the new parameter configuration, replacing the time-consuming calculations based on the physical model.

[0043] For example, in practical applications, numerical optimization algorithms and mapping models can be used to iteratively optimize the magnetic spatial parameters of the magnetic sphere to obtain the objective function values ​​for each round of optimization. The implementation of this scheme is as follows: S21: Gradually adjust the rotation angle of one of the magnetic units and obtain the corresponding objective function value. Search for the rotation angle corresponding to the minimum objective function value and use it as the target angle of the current magnetic unit.

[0044] S22: Fix the magnetic unit with the known target angle to the corresponding state, and use the method of S21 to determine the target angle of each of the remaining magnetic units in sequence; S23: The objective function value with all magnetic units at the target angle is taken as the final optimization result.

[0045] The numerical optimization algorithm described above can employ the Nelder-Mead polyhedron search algorithm. Assuming there are N magnetic units, with a pitch range of (-80°, 80°), discretized into 160 smaller intervals, the rotation angle of magnetic unit 1 is gradually adjusted within this range. Magnetic units 2 through N are not adjusted at this time. The Nelder-Mead polyhedron search algorithm is used in conjunction with the objective function to optimize parameters, finding the rotation angle corresponding to the minimum objective function value as the target angle. The pitch angle of magnetic unit 1 is then fixed to this target angle. The same method is then used to adjust magnetic units 2 through N sequentially. Each time a magnetic unit is adjusted, the rotation angles of the other magnetic units must remain constant. Compared to traditional optimization algorithms, this method employs a parameter decoupling strategy to effectively reduce the problem dimensionality, significantly reducing computational complexity while maintaining optimization accuracy. It is particularly suitable for multi-magnet collaborative control scenarios, demonstrating significant advantages in system response speed and imaging equipment compatibility.

[0046] S3: Treat each data pair as a sample data to obtain a dataset containing a large number of sample data; divide the dataset into training set, test set, and validation set, train the pre-built neural network model until the loss function converges, and obtain the trained neural network model.

[0047] The neural network model can adopt an MPL (Multilayer Perceptron) neural network model. The physical model ensures the correctness of the theory, while the neural network model can accelerate the search process. The two work together to achieve a balance between accuracy and efficiency.

[0048] In summary, this invention provides a magnetically controlled navigation system based on a permanent magnet array. The magnetic array is equipped with a first adjustment mechanism and a second adjustment mechanism to achieve full-angle adjustment of the magnetic ball, thereby improving the magnetic field adjustment capability of the magnetically controlled navigation system. The third adjustment mechanism adjusts the position of the magnetic ball, avoiding the inconvenience caused by moving the robotic arm, further improving the flexibility of magnetic field adjustment of the magnetically controlled navigation system, and also reducing the volume occupied by the magnetically controlled navigation system. Since no electromagnets are used, there is no need to equip it with a cooling system. The optimal distribution mode of the magnetic array can be found using a pre-trained neural network model to improve the magnetic field strength and isotropy. Through step-by-step iterative optimization and neural network assistance, the computational complexity is reduced and the system response speed is improved.

[0049] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. The above descriptions are merely specific embodiments of this invention, but the scope of protection of this invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this invention should be included within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. A magnetically controlled navigation system based on a permanent magnet array, characterized in that, It includes: Two mirror-mounted, height-adjustable support structures; Two movable magnetic arrays are respectively disposed on top of the two support mechanisms; each magnetic array includes a base and multiple magnetic units distributed on its surface, and the magnetic units in the two magnetic arrays are arranged opposite each other along a vertical plane; each magnetic unit includes a permanent magnet-type magnetic ball, a first adjustment mechanism, a second adjustment mechanism, and a third adjustment mechanism; the first adjustment mechanism is used to drive the magnetic ball to rotate in a direction perpendicular to the magnetic axis; the second adjustment mechanism is used to drive the magnetic ball to rotate in a rotation direction perpendicular to the first adjustment mechanism; the third adjustment mechanism is used to drive the magnetic ball to pitch along a vertical plane to adjust the position of the magnetic ball; A magnetic field control module is electrically connected to the magnetic array. The magnetic field control module includes a spatial optimization unit and an execution unit. The spatial optimization unit runs a pre-trained neural network model, which is used to predict the performance indicators of the magnetic field formed by each magnetic sphere in the magnetic array under different pitch states, including magnetic field strength and isotropy. The spatial optimization unit is used to first iteratively optimize the optimal spatial distribution state of the magnetic array by combining the neural network model. Then, under the optimal spatial distribution state of the magnetic array, the rotation angle of each magnetic sphere is solved in reverse by the control algorithm to make the target magnetic field between the two magnetic arrays form the required magnetic field. The execution unit is used to first issue instructions to each third adjustment mechanism to adjust the pitch angle of each magnetic unit so that the magnetic array is in the optimal spatial distribution state. Then, it issues instructions to the first adjustment mechanism and the second adjustment mechanism to adjust the rotation angle of each magnetic unit to obtain the required magnetic field distribution.

2. The magnetically controlled navigation system based on a permanent magnet array according to claim 1, characterized in that: The first adjustment mechanism includes a first rotating shaft, an outer support, and a first driver; the magnetic ball is movably connected to the outer support via a first rotating shaft perpendicular to the magnetic axis of the magnetic ball; the first driver is drively connected to the first rotating shaft; The second adjustment mechanism includes a second rotating shaft, a main support, and a second driver; the outer support is rotatably connected to the main support via a second rotating shaft perpendicular to the first rotating shaft; the second driver is drively connected to the second rotating shaft. The third adjustment mechanism includes a pitch rotation shaft, a pitch driver, and an angle lock; the pitch rotation shaft is fixed to the main support in the horizontal direction and is hinged to the outer support; the pitch driver is fixed to the main support and is connected to the outer support in a transmission manner; the angle lock is located at the end of the pitch rotation shaft and locks the angle of the outer support by increasing friction or mechanically snapping.

3. The magnetically controlled navigation system based on a permanent magnet array according to claim 2, characterized in that, The outer support includes two parallel connecting rods and a bracket connecting the ends of the two connecting rods on the same side; the connecting rods are perpendicular to the first rotation axis and parallel to the second rotation axis; the base of the first driver is fixedly connected to one of the connecting rods, so that the first driver rotates in a direction perpendicular to the two connecting rods; the magnetic ball rotates through the rotation of the first driver, which is connected to the first rotation axis; the bracket is rotatably connected to the main support through a ball bearing.

4. The magnetically controlled navigation system based on a permanent magnet array according to claim 2 or 3, characterized in that, The first driver is connected to the first rotating shaft via a gear assembly; the gear assembly includes a drive wheel, a transmission wheel, and a driven wheel; the transmission wheel includes two layers of gears, the first layer of gears meshing with the drive wheel, and the second layer of gears meshing with the driven wheel; the drive wheel is fixedly connected to the first driver; the driven wheel is fixedly connected to the first rotating shaft.

5. The magnetically controlled navigation system based on a permanent magnet array according to claim 3, characterized in that, The second driver and the second rotating shaft are connected via a support assembly; the support assembly includes: a rotating support vertically connected between the two connecting rods, and a drive support fixedly connected between the second driver and the bracket; the rotating support is fixedly connected to the base of the second driver, so that the second driver rotates in a direction perpendicular to the rotating support; the drive support drives the outer support to rotate through the rotation of the second driver.

6. The magnetically controlled navigation system based on a permanent magnet array according to claim 1, characterized in that, The magnetic array comprises nine magnetic units arranged in a 3×3 configuration. For each column of magnetic units, three magnetic spheres are located on the same circumscribed circle, and the angles between the line connecting the center of the magnetic spheres on both sides to the center of the circumscribed circle and the horizontal line are 57° and -47°, respectively; the angle between the line connecting the center of the magnetic sphere in the middle to the center of the circumscribed circle and the horizontal line is 5°.

7. The magnetically controlled navigation system based on a permanent magnet array according to claim 1, characterized in that, The design method for the neural network model includes: S1: Establish a mapping model, which pre-defines the mapping relationship between the magnet array configuration and the magnetic field characteristics; S2: Design an objective function with magnetic field strength and isotropy as optimization objectives; use numerical optimization algorithms and the mapping model to iteratively optimize the magnetic space parameters of the magnetic sphere to obtain the objective function values ​​for each round of optimization; combine the input magnetic space parameters and the output objective function values ​​in each round of optimization into a data pair; S3: Treat each data pair as a sample data to obtain a dataset containing a large number of sample data; divide the dataset into training set, test set, and validation set, train the pre-built neural network model until the loss function converges, and obtain the trained neural network model.

8. The magnetically controlled navigation system based on a permanent magnet array according to claim 7, characterized in that, Step S2, which involves iteratively optimizing the magnetic space parameters of the magnetic sphere using a numerical optimization algorithm and the mapping model to obtain the objective function values ​​for each round of optimization, includes: S21: Gradually adjust the rotation angle of one of the magnetic units and obtain the corresponding target function value, search for the rotation angle corresponding to the minimum target function value, and take it as the target angle of the current magnetic unit; S22: Fix the magnetic unit with the known target angle to the corresponding state, and use the method of S21 to determine the target angle of each of the remaining magnetic units in sequence; S23: The objective function value with all magnetic units at the target angle is taken as the final optimization result.

9. The magnetically controlled navigation system based on a permanent magnet array according to claim 8, characterized in that, The numerical optimization algorithm is the Nelder-Mead polyhedron search algorithm.

10. The magnetically controlled navigation system based on a permanent magnet array according to claim 7, characterized in that, The neural network model is a multilayer perceptron neural network model.

Images