A programmable topologically universal electromagnetic platform system and control method

Abstract

The present application relates to a programmable topological universal electromagnetic platform system and a control method. The system comprises a programmable distributed electromagnetic matrix, a field topology control module, a digital twin module and an AI prediction control module; a space-time driving matrix containing electromagnetic cell gating state, current size, current direction and driving timing is generated according to the target force vector, and the magnetic field topology is dynamically reconstructed. The digital twin module predicts the magnetic field, force response and constraint state corresponding to the candidate drive, and the AI prediction control module predicts the environmental disturbance and the state of the controlled object and implements feedforward compensation. Single-layer coils output bidirectional force through timing commutation, multi-layer coils are nested and coupled to superimpose power, and coils are multiplexed between driving and energy feedback states. The system is not limited to a single application object.

Description

Technical Field

[0001] This invention relates to the fields of electromagnetic field generation, distributed electromagnetic drive, digital twin and intelligent control technology, specifically to a programmable and topologically adaptable general electromagnetic platform system and control method. Background Technology

[0002] Electromagnetic force control can be applied to various technological scenarios at different scales, including magnetic levitation, precision positioning, aircraft takeoff and landing assistance, microscopic magnetohydrodynamic manipulation, and plasma or charged particle guidance. Existing electromagnetic actuators are typically designed around specific controlled objects and fixed tasks. The coil positions, connections, energizing directions, and control channels are basically fixed after manufacturing. What can be adjusted is mainly the magnetic field strength near a limited operating point, and it is difficult to change the spatial connection relationship, gradient direction, and effective area of ​​the magnetic field through software.

[0003] While some multi-coil systems can adjust the coil current individually, they typically operate according to preset channel combinations or fixed lookup table methods, lacking a topology control mechanism that dynamically selects electromagnetic units based on the target force vector and jointly adjusts the current magnitude, direction, and timing. When the controlled object's position changes, external magnetic field disturbances occur, load changes abruptly, or electromagnetic unit parameters drift, fixed control parameters struggle to continuously output an accurate target force vector.

[0004] Furthermore, multi-element arrays suffer from electromagnetic coupling, magnetic saturation, temperature rise, power supply constraints, and the mutual influence of the controlled object's dynamics. Relying solely on offline finite element models or single feedback control makes it difficult to simultaneously predict the magnetic field topology, force response, energy consumption, and thermal state corresponding to candidate control variables, and also makes it difficult to implement feedforward compensation before the actual disturbance occurs. Traditional coils are also typically used only as energy-consuming actuators, and the energy generated when the controlled object decelerates, descends, or the coil demagnetizes is not effectively recovered.

[0005] Therefore, there is a need for a programmable and topologically adaptable general electromagnetic platform system that is not limited to a single application scenario. This system enables distributed electromagnetic units to be independently selected and controlled by the magnitude, direction, and timing of current. It forms a predictable and correctable magnetic field topology reconstruction closed loop through digital twins and AI predictive control, while also taking into account single-layer bidirectional force output, multi-layer power expansion, and bidirectional electromechanical energy conversion. Summary of the Invention

[0006] According to embodiments of the present invention, a programmable and topologically adaptable universal electromagnetic platform system and a control method are provided to address the problems mentioned in the background section.

[0007] In a first aspect of the invention, a programmable and topologically adaptable universal electromagnetic platform system is provided.

[0008] The programmable and topologyable general-purpose electromagnetic platform system includes a programmable distributed electromagnetic matrix, a field topology control module, a digital twin module, and an AI predictive control module. The programmable distributed electromagnetic matrix includes multiple spatially arranged electromagnetic units that can be independently selected and have their driving currents adjusted. The field topology control module generates a spatiotemporal driving matrix based on a target force vector. This spatiotemporal driving matrix at least determines the participating electromagnetic units and the current magnitude, direction, and driving sequence of each unit. It then controls the programmable distributed electromagnetic matrix to dynamically reconstruct the magnetic field topology, outputting an electromagnetic force corresponding to the target force vector within the workspace. The digital twin module establishes a virtual mapping between the programmable distributed electromagnetic matrix, the controlled object, and the workspace, and predicts the magnetic field topology, force response, and constraint state corresponding to candidate spatiotemporal driving matrices. The AI ​​predictive control module predicts environmental disturbances and the state of the controlled object based on the target force vector, historical spatiotemporal driving matrices, sensor feedback, and the prediction results from the digital twin module. It generates a feedforward compensation amount for the spatiotemporal driving matrix and updates this compensation amount based on execution feedback.

[0009] Preferably, each of the electromagnetic units includes a coil, a gating switch, and a bidirectional current drive interface; the programmable distributed electromagnetic matrix includes a single-layer coil array, and the field topology control module switches the composite magnetic field gradient or composite force component between the attraction direction and the repulsion direction by switching the current direction of at least some of the coils and adjusting the excitation timing of adjacent coils.

[0010] Preferably, the programmable distributed electromagnetic matrix includes at least two layers of coil arrays nested, staggered, or partially overlapping along the thickness direction. The field topology control module performs synchronous excitation, phase-shifting excitation, or hierarchical excitation on different layers of coils to couple and superimpose the electromagnetic forces generated by different layers of coils.

[0011] Preferably, at least a portion of the coils are connected to the energy storage bus via a bidirectional power conversion circuit; the coils convert electrical energy into electromagnetic energy in the driving state, and switch to an energy feedback state when the controlled object releases mechanical energy, the controlled object cuts magnetic lines of force, or the coil demagnetizes, so as to send the induced electrical energy or coil energy back to the energy storage bus.

[0012] Preferably, the digital twin module includes at least one of an array geometric model, a circuit model, a thermal model, an electromagnetic coupling model, a controlled object model, and an environmental disturbance model, and corrects the model parameters online based on at least one of measured current, temperature, field strength, force, position, and attitude; the AI ​​predictive control module includes a time-series state encoder, a spatial correlation encoder, a disturbance predictor, and a control quantity generator, and uses the target force vector, unit state, environmental state, historical spatiotemporal driving matrix, and their corresponding measured force response as training samples, with reducing the error between the predicted force response and the measured force response as at least one training objective.

[0013] Preferably, the controlled object is an aircraft or an aircraft carrier component, and the target force vector is generated based on the aircraft's position, attitude, speed, and desired takeoff trajectory, hovering position, or landing trajectory, and is used to perform at least one of takeoff lift, hovering, braking, lateral correction, attitude adjustment, or terminal position guidance.

[0014] Preferably, the controlled object is a magnetic particle, a magnetic microcarrier, a magnetic droplet, or a magnetic fluid containing magnetic components. The target force vector is generated based on the target position, target velocity, target concentration distribution, or sorting path, and is used to perform at least one of aggregation, sorting, transport, positioning, or mixing.

[0015] Preferably, the controlled object is plasma, ion flow, or charged particle beam, the target force vector is generated based on the target propagation path, convergence position, beam size, or spatial density distribution, and the field topology control module forms a moving, bending, focusing, or diffusing magnetic field topology by changing the electromagnetic units involved in the execution, the current direction, and the driving timing.

[0016] A control method for a programmable and topologyable general-purpose electromagnetic platform system includes: acquiring a target force vector, the unit states of multiple independent controllable electromagnetic units, the state of the controlled object, and the environmental state; generating a candidate spatiotemporal driving matrix based on the target force vector, including at least the electromagnetic unit selection state, current magnitude, current direction, and driving timing; predicting the magnetic field topology, force response, and constraint state corresponding to the multiple candidate spatiotemporal driving matrices by a digital twin module; predicting environmental disturbances and the state of the controlled object by an AI predictive control module, and determining the target spatiotemporal driving matrix or generating a correction amount for the candidate spatiotemporal driving matrix from the candidate spatiotemporal driving matrices based on the target force vector and the prediction results of the digital twin module; independently selecting the electromagnetic units according to the target spatiotemporal driving matrix and adjusting the current magnitude, current direction, and driving timing of each participating electromagnetic unit to dynamically reconstruct the magnetic field topology and output the electromagnetic force corresponding to the target force vector; acquiring execution feedback and continuously correcting the digital twin module, the AI ​​predictive control module, and the target spatiotemporal driving matrix.

[0017] Preferably, the current direction and excitation sequence of at least some coils in the single-layer coil array are switched at adjacent control moments to obtain force components in the attraction or repulsion direction; and / or, at least two layers of coil array are synchronously excited, phase-shifted excited, or graded excited to superimpose electromagnetic forces; and / or, when the controlled object releases mechanical energy or the coil demagnetizes, at least some coils are switched to energy feedback state and power is fed back to the energy storage bus.

[0018] One or more technical solutions provided in this application have at least the following technical effects or advantages: Using distributed, independently controllable electromagnetic units as the unified underlying hardware, the magnetic field topology can be reconstructed for different objects and tasks; single-layer coils obtain bidirectional force output through timing commutation, while multi-layer coils extend power and depth of action through nested coupling; digital twins and AI predictive control can generate feedforward compensation before disturbances occur and achieve adaptive updates based on execution feedback; coils can be reused between driving and feedback states to improve energy utilization; each application scenario only changes the target force vector generation method, the controlled object model, and the constraints, without changing the underlying architecture of the general system.

[0019] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 This is a schematic diagram of the general architecture and closed-loop control relationship of the programmable and topologically adaptable universal electromagnetic platform system of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0023] The target force vector in this invention describes the direction and magnitude of the desired electromagnetic force within the workspace, and may also include the force gradient, target position, effective time window, and rate of change. For continuous spatial targets, the workspace can be discretized into multiple control points, and target force vectors can be set for each. The system solves for the corresponding spatiotemporal driving matrix under constraints of current, bus voltage, temperature rise, magnetic field safety boundary, and the controlled object's motion range.

[0024] Example 1: General Architecture and Intelligent Force Field Control The system in this embodiment includes a programmable distributed electromagnetic matrix, a field topology control module, a digital twin module, an AI predictive control module, a sensor feedback link, and an optional bidirectional energy interface. The controlled object can be located on one side of the array, within the space enclosed by the array, or between multiple arrays. The sensor feedback collects at least one of the following: current, temperature, field strength, force, position, and attitude. Different applications may also collect velocity, concentration, flow field, or particle distribution. In manual physical verification mode, the operator manually selects the electromagnetic units involved in the execution via a host computer, programmable power supply, or control switch, and sets the current magnitude, current direction, phase, and on / off sequence. The field topology control module converts the manual settings into a spatiotemporal driving matrix and outputs it to each electromagnetic unit to physically verify the formed magnetic field topology and electromagnetic force.

[0025] The programmable distributed electromagnetic matrix consists of M independently controllable electromagnetic units. The control quantity of the i-th electromagnetic unit can be expressed as ui(t)=[si(t), Ii(t), di(t), φi(t), pi(t)], where si represents the gating state, Ii represents the current amplitude, di represents the current direction, φi represents the phase, and pi represents the duty cycle or pulse width. The control quantities of all electromagnetic units are combined in spatial position and time order to form a spatiotemporal driving matrix U(t). In the manual engineering platform mode, a control parameter library is established according to the physically verified electromagnetic unit combination, driving parameters, and calibration results. The digital twin module uses the control parameter library to establish an initial model, and the AI ​​predictive control module uses manual experimental data and operational feedback for training and updating, enabling the system to be extended to automatic predictive control while retaining manual setting and human intervention functions.

[0026] Neglecting magnetic saturation or performing piecewise corrections for it, the synthetic magnetic flux density at position r in the workspace can be expressed as B(r,t)=ΣGi(r,x,T)·Ii(t), where Gi represents the influence function of the i-th electromagnetic unit on the magnetic field at position r under the controlled object state x and temperature T. For nonlinear magnetic materials, the digital twin module corrects Gi using hysteresis curves, finite element lookup tables, or online identification models. The field topology control module changes the extreme values, zero points, gradient directions, and connected regions of the synthetic magnetic field by altering the gating set and the magnitude, direction, and timing of Ii(t).

[0027] For a single-layer coil array, the field topology control module switches the current direction of the selected coil at adjacent control moments and adjusts the excitation sequence of adjacent units, causing the sign of the resultant magnetic field gradient in the selected axial or tangential direction to change. For objects with a known magnetic moment, gradient reversal causes the force to switch between the attractive and repulsive directions; for inductive objects, bidirectional resultant force components are generated through the combination of multi-unit phase difference and spatial gradient.

[0028] When a task requires greater peak force, depth of action, or fault tolerance, two or more layers of coil arrays are arranged in the thickness direction. Different layers employ coaxial nesting, staggered overlapping, or combinations of coils of different sizes. The field topology control module selects synchronous excitation, phase-shifting excitation, or hierarchical excitation based on the target force vector. Minor control quantities are provided by small-sized coils, while high-power base quantities are provided by large-sized coils or multiple layers of synchronous excitation.

[0029] The bidirectional energy interface includes a bidirectional power conversion circuit connecting the coil and the energy storage bus. When the electromagnetic unit outputs force, energy flows from the energy storage bus to the coil; when the controlled object descends, decelerates, or is driven by an external force, the corresponding coil is connected to the bidirectional power conversion circuit according to the feedback sequence, and the induced electrical energy is rectified and voltage-converted before being sent back to the energy storage bus; the magnetic field energy released when the coil demagnetizes is also recovered. The recovered energy Erec can be expressed as Erec = ∫Prec(t)dt, where Prec is the instantaneous feedback power received by the energy storage bus.

[0030] The digital twin module includes an array geometry model, a circuit model, a thermal model, an electromagnetic coupling model, a controlled object dynamics model, and an environmental disturbance model. The array geometry model records the position, axis, and effective operating area of ​​the electromagnetic units; the circuit and thermal models record resistance, inductance, drive delay, temperature rise, and heat dissipation; the controlled object model records mass, magnetic moment, permeability, fluid parameters, charge density, or other parameters related to the electromagnetic force response, depending on the object type. The digital twin module takes the candidate spatiotemporal drive matrix as input and outputs the predicted magnetic field topology, predicted force vector, predicted object state, predicted temperature rise, and predicted energy consumption.

[0031] One implementation of the AI ​​predictive control module includes a temporal state encoder, a spatial correlation encoder, a disturbance predictor, and a control quantity generator. The spatial correlation encoder treats electromagnetic units as graph nodes and the electromagnetic coupling relationships between units as graph edges; the temporal state encoder processes historical spatiotemporal driving matrices and sensor feedback; the disturbance predictor outputs the predicted environmental disturbances and object states in the time domain; and the control quantity generator outputs candidate spatiotemporal driving matrices or their corrections. Alternatively, temporal networks, graph neural networks, reinforcement learning policy networks, or hybrid physical information networks with the same input-output relationships can also be used.

[0032] The model training samples include at least the target force vector, element states, environmental states, spatiotemporal driving matrices, and the measured field strength, force, and object states after execution. Before training, continuous variables are normalized, and perturbations are applied to mitigate electromagnetic element failure, external magnetic field changes, load abrupt changes, and sensor noise. Training objectives include at least reducing the error between the predicted and measured force responses, and may also include penalties for temperature rise prediction errors, object state prediction errors, and energy consumption. The trained AI model is jointly deployed with a digital twin physical model. The digital twin is used to constrain the physically feasible region, while the AI ​​model is used to compensate for perturbations and parameter biases that are difficult to model precisely.

[0033] During online control, the AI ​​predictive control module solves for the control quantity within a prediction time domain of length H. The objective function can be expressed as J(U) = Σ||Fref(k) - F̂(k|U)|| 2 The formula is: +λE·E(U)+λT·Ptemp(U)+λS·ΔU, where Fref is the target force vector, F̂ is the force vector jointly predicted by the digital twin and AI model, E is the energy consumption, Ptemp is the penalty term for exceeding the temperature rise limit, ΔU is the change in the driving matrix between adjacent control cycles, and λE, λT, and λS are weights. The obtained control quantity U*=argmin J(U) satisfies the constraints of coil current, bus voltage, temperature rise, magnetic field safety boundary, and object motion range. The first control step of U* is executed in each control cycle, and the state is updated according to the new sensor feedback and the solution is recalculated.

[0034] The general control flow includes: receiving the target force vector, task constraints, and sensor feedback; updating the array, object, and environmental states in the digital twin; predicting the magnetic field topology, force response, temperature rise, and energy consumption corresponding to multiple candidate spatiotemporal driving matrices; predicting disturbances and generating feedforward compensation by the AI ​​predictive control module; gating electromagnetic units according to the target spatiotemporal driving matrix and adjusting the current magnitude, direction, and timing; acquiring the actual response and correcting the digital twin parameters and AI predictive control parameters online. This process is executed continuously throughout the task. If an electromagnetic unit fails or overheats, the corresponding unit is removed from the gating set, and the remaining units are used to resolve the problem.

[0035] Example 2: Aircraft Takeoff and Landing and High-Gload Control In aircraft takeoff and landing applications, the aircraft transmits target force information to the platform via a wireless data link. The platform uses the programmable distributed electromagnetic matrix described in Example 1 as its execution layer. External safety monitoring is only used to view the operational status and does not participate in the main closed-loop control. The aircraft can be configured with an IMU, a downward-looking vision sensor, millimeter-wave radar, and a magnetometer to fuse and estimate the aircraft's position, attitude, and velocity relative to the platform, and generate a time-varying target force vector based on the desired takeoff trajectory, hovering position, or landing trajectory.

[0036] During takeoff, the multi-layer coil array first establishes a baseline lift force to bear the main load, and then small-sized coils output differential torque or lateral correction force based on attitude deviations. During hovering or trajectory tracking, the digital twin predicts the load capacity, attitude changes, and temperature rise under different candidate drive matrices, and the AI ​​predictive control module generates feedforward compensation based on crosswinds, load changes, and descent speed. As the aircraft approaches the platform end, the array outputs the composite force vector required for lift, braking, attitude adjustment, and terminal position guidance.

[0037] The potential and kinetic energy released during the aircraft's descent or braking phase is converted into induced electrical energy through electromagnetic coupling between the aircraft and the coil. The bidirectional power conversion circuit then feeds this energy back to the energy storage bus. If a single electromagnetic unit overheats or fails, the field topology control module removes it from the gating set and redistributes the force vector using adjacent units to avoid sudden changes in local forces.

[0038] Example 3: Manipulation of Microscopic Magnetofluids and Magnetic Particles In microscopic manipulation scenarios, the controlled objects are magnetic particles, magnetic microcarriers, magnetic droplets, or magnetofluids containing magnetic components. The workspace can be a microchannel, reaction chamber, or chip surface. The target force vector is generated by the target position, target velocity, target concentration distribution, or sorting path. Sensing feedback can include microscopic images, flow rates, particle positions, and concentration distributions.

[0039] The array preferentially employs single-layer small-sized coils or locally multi-layer nested coils. The field topology control module sequentially selects adjacent electromagnetic units to form a moving magnetomotive force trap. By changing the lateral force component through current commutation, and by adjusting the local gradient, it achieves aggregation, sorting, transport, positioning, or mixing. The object model in the digital twin records particle magnetic moments, fluid viscosity, flow channel resistance, and Brownian perturbation parameters. The AI ​​predictive control module predicts particle displacement based on historical images and flow field changes, and pre-corrects the position and velocity of the moving magnetomotive force trap.

[0040] The microscopic task aims for high spatial resolution and low force fluctuations, employing low current and a high update frequency. When it is necessary to expand the depth of action, only the multilayer coils corresponding to the target region are excited. This implementation still follows the target force vector generation, digital twin prediction, AI feedforward compensation, and spatiotemporal drive matrix control process described in Example 1.

[0041] Example 4: Plasma, Ion Stream, and Charged Particle Guidance In plasma or charged particle guided scenarios, the controlled object is plasma, ion flow, or charged particle beam. The target force vector is generated based on the desired propagation path, convergence location, beam spot size, or spatial density distribution. Sensing feedback can include magnetic field probe signals, spectra, imaging results, beam position, and current density.

[0042] The controlled object model in the digital twin uses the equations of motion of charged particles in an electromagnetic field, along with plasma density, velocity, collision frequency, and boundary conditions, to predict the trajectory and spatial distribution corresponding to candidate magnetic field topologies. The AI ​​predictive control module generates feedforward compensation based on changes in the inlet state, external magnetic field disturbances, and cavity boundary changes. The field topology control module, by changing the electromagnetic units involved in the execution, current direction, phase, and timing, forms a moving, bending, focusing, or spreading magnetic field topology in the propagation direction, guiding the controlled object to the target region.

[0043] The aforementioned aircraft, magnetohydrodynamic, and plasma embodiments only change the method of generating the target force vector, the controlled object model, and the constraints; they do not alter the general underlying architecture comprised of the programmable distributed electromagnetic matrix, spatiotemporal driving matrix, magnetic field topology reconstruction, digital twin prediction, and AI predictive control. Those skilled in the art can modify the shape, quantity, spatial arrangement, number of coil layers, and power level of the electromagnetic units according to the workspace shape, target force range, and characteristics of the controlled object; any equivalent substitutions made still fall within the inventive concept.

[0044] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A programmable and topologyable universal electromagnetic platform system, characterized in that, It includes a programmable distributed electromagnetic matrix, a field topology control module, a digital twin module, and an AI predictive control module. The programmable distributed electromagnetic matrix includes multiple spatially arranged electromagnetic units that can be independently selected and have their driving currents independently adjusted. The field topology control module is used to generate a spatiotemporal driving matrix based on a target force vector. The spatiotemporal driving matrix at least determines the electromagnetic units involved in the execution, as well as the current magnitude, current direction, and driving sequence of each participating electromagnetic unit. It also controls the programmable distributed electromagnetic matrix to dynamically reconstruct the magnetic field topology according to the spatiotemporal driving matrix, so as to output an electromagnetic force corresponding to the target force vector within the workspace. The digital twin module is used to establish a virtual mapping between the programmable distributed electromagnetic matrix, the controlled object, and the workspace, and to predict the magnetic field topology, force response, and constraint state corresponding to the candidate spatiotemporal driving matrix; the AI ​​prediction control module is used to predict environmental disturbances and the state of the controlled object based on the target force vector, the historical spatiotemporal driving matrix, sensor feedback, and the prediction results of the digital twin module, to generate the feedforward compensation amount of the spatiotemporal driving matrix, and to update the feedforward compensation amount based on the execution feedback.

2. The programmable and topologyable universal electromagnetic platform system according to claim 1, characterized in that, Each of the electromagnetic units includes a coil, a gating switch, and a bidirectional current drive interface; the programmable distributed electromagnetic matrix includes a single-layer coil array, and the field topology control module switches the composite magnetic field gradient or composite force component between the attraction direction and the repulsion direction by switching the current direction of at least some of the coils and adjusting the excitation timing of adjacent coils.

3. The programmable and topologyable universal electromagnetic platform system according to claim 2, characterized in that, The programmable distributed electromagnetic matrix includes at least two layers of coil arrays nested, staggered, or partially overlapping along the thickness direction. The field topology control module performs synchronous excitation, phase-shifting excitation, or hierarchical excitation on different layers of coils to couple and superimpose the electromagnetic forces generated by different layers of coils.

4. The programmable and topologyable universal electromagnetic platform system according to claim 2, characterized in that, At least a portion of the coils are connected to the energy storage bus via a bidirectional power conversion circuit; the coils convert electrical energy into electromagnetic energy in the driving state, and switch to energy feedback state when the controlled object releases mechanical energy, the controlled object cuts magnetic lines of force, or the coil demagnetizes, so as to send the induced electrical energy or the coil stored energy back to the energy storage bus.

5. The programmable and topologyable universal electromagnetic platform system according to claim 1, characterized in that, The digital twin module includes at least one of an array geometric model, a circuit model, a thermal model, an electromagnetic coupling model, a controlled object model, and an environmental disturbance model, and corrects the model parameters online based on at least one of measured current, temperature, field strength, force, position, and attitude. The AI ​​predictive control module includes a time-series state encoder, a spatial correlation encoder, a disturbance predictor, and a control quantity generator, and uses the target force vector, unit state, environmental state, historical spatiotemporal driving matrix, and their corresponding measured force response as training samples, with reducing the error between the predicted force response and the measured force response as at least one training objective.

6. The programmable and topologyable universal electromagnetic platform system according to claim 1, characterized in that, The controlled object is an aircraft or an aircraft carrier component. The target force vector is generated based on the aircraft's position, attitude, speed, and desired takeoff trajectory, hovering position, or landing trajectory, and is used to perform at least one of takeoff lift, hovering, braking, lateral correction, attitude adjustment, or terminal position guidance.

7. The programmable and topologyable universal electromagnetic platform system according to claim 1, characterized in that, The controlled object is a magnetic particle, a magnetic microcarrier, a magnetic droplet, or a magnetic fluid containing magnetic components. The target force vector is generated based on the target position, target velocity, target concentration distribution, or sorting path, and is used to perform at least one of aggregation, sorting, transport, positioning, or mixing.

8. The programmable and topologyable universal electromagnetic platform system according to claim 1, characterized in that, The controlled object is plasma, ion flow, or charged particle beam. The target force vector is generated based on the target propagation path, convergence position, beam spot size, or spatial density distribution. The field topology control module forms a moving, bending, focusing, or diffusing magnetic field topology by changing the electromagnetic units involved in the execution, the current direction, and the driving timing.

9. A control method for a programmable and topologyable universal electromagnetic platform system, characterized in that, include: Acquire the target force vector, the unit state of multiple independent controllable electromagnetic units, the state of the controlled object, and the environmental state; Based on the target force vector, a candidate spatiotemporal driving matrix is ​​generated, which includes at least the electromagnetic unit gating state, current magnitude, current direction, and driving timing. The digital twin module predicts the magnetic field topology, force response, temperature rise, energy consumption, and constraint state corresponding to multiple candidate spatiotemporal driving matrices; the AI ​​predictive control module solves for the control quantity within a prediction time domain of length H, expressed as J(U)=Σ||Fref(k)-F̂(k|U)|| 2 The objective function is +λE·E(U)+λT·Ptemp(U)+λS·ΔU, where Fref is the target force vector, F̂ is the force vector jointly predicted by the digital twin module and the AI ​​predictive control module, E(U) is the energy consumption, Ptemp(U) is the penalty term for exceeding the temperature rise limit, ΔU is the change in the spatiotemporal driving matrix between adjacent control cycles, and λE, λT and λS are weights. Under the conditions of satisfying the constraints of coil current, bus voltage, temperature rise, magnetic field safety boundary and the motion range of the controlled object, the control quantity U*=argmin J(U) is obtained, and the target spatiotemporal driving matrix or the correction amount of the candidate spatiotemporal driving matrix is ​​determined based on the control quantity. The electromagnetic units are independently selected according to the target spatiotemporal driving matrix, and the current magnitude, current direction and driving timing of each participating electromagnetic unit are adjusted to dynamically reconstruct the magnetic field topology and output the electromagnetic force corresponding to the target force vector. In each control cycle, the first control step corresponding to the control quantity U* is executed to obtain execution feedback, and the digital twin module, the AI ​​predictive control module and the target spatiotemporal driving matrix are continuously corrected.

10. The control method according to claim 9, characterized in that, Switch the current direction and excitation sequence of at least some coils in the single-layer coil array at adjacent control moments to obtain force components in the attraction or repulsion direction; And / or, synchronously excite, out-of-phase excite, or graded excite at least two layers of coil array to superimpose electromagnetic forces; and / or, when the controlled object releases mechanical energy or the coils demagnetize, switch at least a portion of the coils to an energy feedback state and send electrical energy back to the energy storage bus.

Images