Millimeter wave FPC multi-domain adaptive tuning antenna and tuning method thereof

Abstract

The invention relates to the technical field of antennas, in particular to a millimeter wave FPC multi-domain self-adaptive tuning antenna and a tuning method thereof, and the method comprises the following steps: S1, constructing an antenna initial topology and frequency model; s2, generating a bionic structure by adopting a topological self-growing algorithm based on the power density distribution; s3, a tuning material with a physical response characteristic is embedded, and impedance and frequency are adjusted by externally stimulating the tuning material; s4, arranging a liquid metal microfluidic channel in the key area, and dynamically reconstructing a conduction path; s5, combining AI image recognition and near field detection to obtain antenna state data; and S6, inputting the acquired antenna state data to an optimization control unit for closed-loop tuning. According to the millimeter wave FPC antenna, structure self-adaption and frequency self-tuning of the millimeter wave FPC antenna in a complex environment are achieved, and the problem that a traditional antenna is unstable in performance in disturbance and frequency change scenes is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of antenna technology, and in particular to a millimeter-wave FPC multi-domain adaptive tuning antenna and its tuning method. Background Technology

[0002] With the continuous evolution of 5G, 6G and millimeter-wave communication technologies, antennas, as key transmitting and receiving components, face higher requirements for performance stability, bandwidth, frequency tuning capability, and environmental adaptability. Traditional microstrip antenna structures typically employ fixed metal patterns and material systems, lacking programmable tuning capabilities and struggling to maintain excellent performance under complex operating conditions such as structural disturbances, frequency band switching, or temperature and humidity changes.

[0003] In recent years, smart antenna structures have also been developed with active tuning modules, but there is still a lack of systematic tuning methods to maintain excellent performance under the aforementioned complex operating conditions, so as to achieve true multi-domain physical coupling and dynamic intelligent control. Therefore, it is necessary to improve them. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by providing a millimeter-wave FPC multi-domain adaptive tuning antenna and its tuning method. This method integrates topology self-evolution, responsive tuning materials, liquid metal conduction reconfiguration, AI image recognition, and near-field feedback control. It can dynamically sense the antenna state and perform closed-loop optimized tuning under complex operating conditions such as structural disturbances, frequency band switching, and environmental changes, thereby significantly improving the frequency tuning capability, matching stability, and environmental adaptability of the millimeter-wave FPC antenna, ensuring its continued excellent performance in variable operating environments.

[0005] To achieve the above objectives, the present invention provides a tuning method for a millimeter-wave FPC multi-domain adaptive tuning antenna, comprising the following steps: S1. Construct the initial topology and frequency target model of the microstrip antenna structure; S2. Using the topology self-growth algorithm, the structure evolves based on the power density distribution to generate an antenna pattern with a biomimetic shape. S3. Embed a tuning material with physical response characteristics in the microstrip antenna structure. Under the action of external physical stimuli, the tuning material itself will change its structure or conductivity to adjust the local matching state, input impedance or resonant frequency of the antenna. S4. Liquid metal microfluidic channels are set in key areas of the microstrip antenna structure. The channels are filled with liquid metal, and the distribution position and shape of the liquid metal in the channels are controlled to change the equivalent conduction path of the microstrip antenna. S5. An AI image recognition module and a near-field electromagnetic detection module are set on the surface of the microstrip antenna. The AI ​​image recognition module is used to identify the geometric deviation image of the microstrip antenna, and the near-field electromagnetic detection module is used to collect spectral response characteristic parameters. S6. The image recognition results and near-field response parameters are jointly input to the optimization control unit, and dynamic parameter matching and tuning command output are executed based on the historical tuning database to complete the automatic structural optimization and matching tuning of the microstrip antenna in the millimeter-wave band.

[0006] A millimeter-wave FPC multi-domain adaptive tuning antenna is also provided for implementing the tuning method described above, the antenna comprising: The flexible microstrip antenna structure features a biomimetic feed pattern optimized through topological self-growth evolution to meet the radiation characteristics of the target frequency band. A tuning material unit is disposed in the antenna structure. The tuning material is responsive to external physical stimuli in order to adjust local structural or electrical performance parameters. A liquid metal microfluidic tuning unit includes at least one closed channel made of a flexible material and liquid metal filled in the closed channel. The channel is located in a critical area of ​​the antenna and can achieve controllable migration of the metal fluid under electrostatic or thermal actuation to dynamically adjust the conduction path. The AI ​​image recognition module and the near-field electromagnetic detection module are used to collect antenna geometric deviation images and spectral response data, respectively. The optimization control unit is used to fuse and analyze the image recognition results and near-field response parameters, and output tuning control signals based on the tuning database to complete the adaptive optimization of the antenna structure and frequency matching tuning.

[0007] The beneficial effects of this invention are as follows: This invention provides a multi-domain adaptive tuning method that integrates topological structure self-evolution, responsive tuning materials, liquid metal conduction reconstruction, AI image recognition, and near-field feedback control. It can dynamically sense the antenna status and perform closed-loop optimized tuning under complex operating conditions such as structural disturbances, frequency band switching, and environmental changes, thereby significantly improving the frequency tuning capability, matching stability, and environmental adaptability of millimeter-wave FPC antennas, ensuring that they maintain excellent performance in variable operating environments. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the steps of a tuning method for a millimeter-wave FPC multi-domain adaptive tuning antenna according to the present invention. Detailed Implementation

[0009] The present invention will now be described in detail with reference to the accompanying drawings.

[0010] like Figure 1As shown, a tuning method for a millimeter-wave FPC multi-domain adaptive tuning antenna according to the present invention includes the following steps: S1. Construct the initial topology and frequency target model of the microstrip antenna structure; ensure that the antenna design has a clear target orientation, provide initial boundary conditions and performance benchmarks for subsequent topology evolution and multi-domain tuning, and improve design efficiency and controllability.

[0011] S2. Using the topology self-growth algorithm, the structure evolves based on the power density distribution to generate antenna patterns with biomimetic morphology; generating irregular but highly adapted antenna patterns optimizes the electromagnetic field distribution and radiation efficiency, thereby improving the antenna's bandwidth and gain performance.

[0012] S3. Embedding a tuning material with physical response characteristics into the microstrip antenna structure. The tuning material includes at least one electrical response material, photodeformable material, thermal conductivity modulating material, or phase change material. It is used to cause changes in the structure or conductivity of the tuning material itself under the action of external physical stimuli, so as to adjust the local matching state, input impedance, or resonant frequency of the antenna, and realize the tuning of electrical parameters caused by microscale structural changes. The tuning of electrical parameters includes micro-adjustment of input impedance, resonant frequency, or reflection coefficient. It facilitates the adaptive response of the antenna structure to environmental changes (such as temperature and humidity, frequency band switching, etc.), and enhances its frequency tunability, impedance matching capability, and environmental stability.

[0013] S4. Liquid metal microfluidic channels are set in key areas of the microstrip antenna structure. The channels are filled with liquid metal. The distribution position and shape of the liquid metal in the channels are actively controlled by the driving unit, thereby changing the equivalent conduction path of the microstrip antenna. This enables programmable reconfiguration of the antenna feed path or electrical length, achieving dynamic tuning at the structural level and macro scale. Without changing the overall antenna structure, fast and reversible frequency tuning and matching path switching are achieved, enhancing the system's adaptability in dynamic communication scenarios.

[0014] S5. An AI image recognition module and a near-field electromagnetic detection module are set on the surface of the microstrip antenna. The AI ​​image recognition module is used to identify the geometric deviation image of the microstrip antenna, and the near-field electromagnetic detection module is used to collect spectral response characteristic parameters. This enables visualization and quantitative monitoring of the antenna status, providing real-time data support for subsequent optimization and tuning, and improving the accuracy and intelligence of the system operation.

[0015] S6. The image recognition results and near-field response parameters are jointly input to the optimization control unit, and dynamic parameter matching and tuning command output are performed based on the historical tuning database to complete the automatic structural optimization and matching tuning of the microstrip antenna in the millimeter-wave band. This realizes closed-loop adaptive optimization tuning control of the millimeter-wave antenna under complex operating conditions, improving the stability, robustness and automation level of the communication system.

[0016] During operation, the initial topological profile of the microstrip antenna and the required operating frequency model are preset as the basic reference structure for optimization and evolution. A topology self-growth algorithm simulates the natural growth process, guiding the conductive pattern to grow gradually based on the electromagnetic power density distribution, forming a high-performance biomimetic structure. Responsive tuning materials (such as thermodeformable materials and electrochromic materials) can undergo reversible changes in physical morphology or electrical properties under external stimuli such as temperature, electric field, or stress. Liquid metals possess excellent conductivity and fluidity; by controlling their flow and positioning within microfluidic channels, the antenna's conduction path and resonant characteristics can be dynamically altered. An AI image recognition module is used to acquire and analyze antenna surface images in real time, identifying structural deviations, deformations, or loss areas; a near-field electromagnetic detection module is used to measure the actual spectral response of the antenna under different structural states (such as reflection coefficient and radiation pattern) in real time. By inputting multi-source data into a deep learning model or control algorithm, comparing and learning from a historical tuning database, the optimal tuning scheme is output, and the antenna structure is adjusted in real time.

[0017] The topology self-growth algorithm in this embodiment establishes a finite element analysis model based on power density distribution, uses Bezier curves or spline functions to describe the shape of the antenna boundary, and performs thickening, branching or shrinking operations on the edge contour of the microstrip antenna structure in each iteration, thereby forming a biomimetic leaf vein-like topology and realizing the generation of distributed feed paths.

[0018] Specifically, by analyzing the energy distribution (power density) on the antenna surface through electromagnetic field simulation, a mathematical model is constructed to quantify performance bottlenecks. This accurately locates weak or inefficient areas of the antenna structure, providing a data foundation for dynamic tuning and ensuring highly targeted optimization.

[0019] Parametric curves (such as Bezier curves and splines) are used to flexibly define the geometry of the antenna edge profile, supporting smooth and continuous shape changes. This enables the antenna structure to achieve high degrees of freedom in deformation, adapting to requirements such as frequency band switching and environmental disturbances, and avoiding the mechanical limitations of traditional fixed structures.

[0020] Thickening, branching, or shrinking operations are performed on the edge contour. Thickening increases the conductor cross-sectional area in high current density regions, reducing ohmic losses; branching expands the current conduction path, forming a multi-branch distributed feed; shrinking reduces the conductor volume in inefficient regions, alleviating ineffective loads. Dynamically optimizing current distribution improves bandwidth and efficiency; generating a biomimetic leaf vein-like topology enhances structural robustness; and realizing distributed feed paths improves impedance matching stability.

[0021] Mimicking the fractal growth mechanism of plant leaf veins, a multi-level branched conductive network is constructed through iterative operations. This enhances environmental adaptability (such as resistance to deformation and temperature and humidity changes), strengthens multi-band resonance capability, and supports wideband tuning.

[0022] By branching, multiple parallel feed channels are naturally extended, generating distributed feed paths to replace traditional single-point feeds. This reduces feed point sensitivity, decreases frequency offset, improves antenna aperture efficiency, and enhances radiation pattern stability.

[0023] Example of implementation module: Hardware control module implementation instructions To achieve dynamic reconfiguration of the antenna structure, the hardware system needs to include two types of core units: 1. Programmable switch arrays (such as PIN diode matrices): By controlling the on / off state of the switching elements, the physical topology of the microstrip antenna metal patch can be directly changed. Specific operations include: Parallel conductors achieve increased thickness (enhancing the conductive cross-sectional area in high current density regions). Extending the branch line enables branching (expanding the distributed power supply path); Disconnecting redundant areas achieves contraction (eliminating inefficient structural loads).

[0024] 2. Adjustable capacitor / inductor arrays (such as MEMS varactors): As compensation units, they dynamically adjust the capacitance or inductance values ​​of the resonant circuit after structural deformation caused by the switching array. For example: The high-frequency resonant point shift caused by branching operation is offset by MEMS varactor voltage tuning. The parasitic parameters introduced by the thickened region are balanced by the adjustable inductance. The two types of modules work together to form a two-dimensional tuning capability of "structure-circuit": the switch array performs macroscopic topology growth, and the reactance array fine-tunes the electromagnetic performance.

[0025] Algorithm processing module implementation instructions Real-time decision-making at the algorithm layer relies on the following computational units: 1. Embedded FPGA / Microcontroller: As the core processor, it runs the topology self-growth algorithm. Solve the finite element model based on the power density distribution to locate the antenna region that needs optimization; Generate Bezier curve control point coordinates to drive the switch array to perform thickening / branching / shrinking operations; 2. Electromagnetic simulation accelerators (such as GPU parallel computing units): These enable high-speed electromagnetic field simulation for structural deformation during iterations. The parallel architecture of GPUs is used to accelerate the solution of Maxwell's equations and predict the radiation properties of the new structure. The simulation results are fed back to the FPGA to optimize the parameters for the next iteration in a closed loop. The two constitute a "decision-verification" cycle, where the FPGA generates tuning instructions and the GPU verifies the performance in real time, ensuring the accuracy of dynamic control.

[0026] The tuning material with physical response characteristics in this embodiment induces changes in conductivity, dielectric constant, structural size deformation, or phase transition behavior through external physical stimuli, thereby adjusting the feed path, electrical length, input impedance, resonant frequency, or return loss of the microstrip antenna structure, and possesses dual self-tuning capabilities in terms of electrical performance and structural morphology.

[0027] By directly responding to physical stimuli such as temperature, humidity, force, and light through materials, the antenna actively compensates for environmental disturbances (such as frequency deviation caused by temperature rise), enabling the antenna to adapt to the environment.

[0028] Dynamic broadband tuning is achieved by switching the resonant frequency in real time through phase transitions or deformations (e.g., VO2 phase transition covers Sub-6G to millimeter waves).

[0029] By simultaneously altering conductivity (impedance matching) and size (electrical length) with a single stimulus, tuning steps are reduced, enabling coordinated optimization of the structure and circuit.

[0030] By eliminating the need for external active circuitry and relying on the intrinsic response of materials, power consumption is reduced by more than 60% compared to traditional electrically tunable antennas, thus improving energy efficiency.

[0031] Implementation example: thermo-electric self-tuning unit. This module employs a miniature heating resistor array and a temperature sensor (such as the DS18B20) to form a temperature control system, combined with a reflected power detection circuit (such as the AD8313 logarithmic detector) to achieve closed-loop control. Its workflow is as follows: the temperature sensor detects the ambient temperature rise in real time, triggering the heating resistor to locally heat the VO2 thin film to the 68°C phase transition point; the VO2 phase transition causes a significant increase in conductivity, leading to a decrease in the antenna resonant frequency; the detector continuously monitors changes in return loss and dynamically adjusts the heating power through a PID algorithm to lock the resonant frequency at the target value, effectively compensating for the -0.38MHz / °C temperature drift effect.

[0032] Implementation example: Electrical-structural tuning unit. This module constructs an electrically excited system based on an interdigital electrode array (gold electrodes, 10μm linewidth) and a high-voltage DAC (such as AD5667, output 0-200V). The workflow is as follows: an embedded algorithm calculates the required electrical length based on the target frequency band, driving the DAC to output a 1MHz / 150V AC voltage to the interdigital electrodes; this electric field stimulates the fluorinated graphene composite film to undergo a 4.2% shrinkage deformation, shortening the microstrip antenna feed path; the reduced electrical length directly increases the resonant frequency by approximately 1.2GHz, achieving millisecond-level switching in the 24.25-27.5GHz millimeter-wave band, meeting the frequency hopping requirements of 5G / 6G.

[0033] Specifically, tuned materials with physical response characteristics include at least one of the following types: 1. The photoresponsive material is a composite film of polyethylene terephthalate liquid crystal and photopolymer, with a response wavelength range of 450 to 700 nm. Irradiation with light at 450-700 nm wavelength triggers changes in the orientation of liquid crystal molecules, altering the material's dielectric constant. The photopolymer's cross-linking / de-cross-linking reaction induces micro-deformation of the structure. Non-contact tuning is achieved by remotely controlling the antenna impedance and resonant frequency through laser irradiation; selective irradiation of specific areas optimizes the radiation pattern, enabling precise local control.

[0034] 2. The thermally responsive material is a vanadium oxide (VO2) thin film that undergoes a metal-insulator phase transition at approximately 68°C; a metal-insulator phase transition also occurs at 68°C, resulting in a sudden change in conductivity (>10). 4 (multiple changes) Phase transitions are accompanied by lattice expansion / contraction, resulting in microscale deformation. Temperature adaptation is achieved by automatically lowering the resonant frequency (increasing the dielectric constant) as ambient temperature rises; simultaneous changes in conductivity (electrical properties) and deformation (structural morphology) enable dual tuning.

[0035] 3. The electroresponsive material is a fluorine-doped graphene composite electrostrictive polymer with an electric field response frequency range of 10 kHz to 1 MHz. Driving the graphene sheets with an electric field of 10 kHz to 1 MHz induces material stretching and deformation. Fluorine doping enhances the polar molecular response, and the dielectric constant varies with the electric field strength gradient. Microsecond-level response to feed path length and input impedance enables high-speed dynamic tuning; significant deformation is achieved with only a weak electric field (<5 V / μm), enabling low-power control.

[0036] 4. Phase change material, a germanium-antimony-tellurium (Ge2Sb2Te5) composite material, is used to construct tunable conductive structures. Thermal / electrical stimulation triggers a phase transition from amorphous to crystalline states, resulting in a dramatic change in conductivity (resistance difference > 10 ohms). 6 (Times); the phase change volume expansion rate is about 6%, resulting in controllable structural deformation.

[0037] By switching the distributed power supply path on and off in the phase transition region, the power supply network can be reconfigured; the amorphous high-resistivity state supports high frequency, while the crystalline low-resistivity state is adapted to low frequency, providing wide bandwidth coverage.

[0038] The liquid metal microfluidic channel in this embodiment is a sealed structure made of flexible polydimethylsiloxane (PDMS). The channel cross-sectional width is 100 micrometers to 500 micrometers. The liquid metal filled inside the liquid metal microfluidic channel is a gallium-indium-tin alloy. The flow of the gallium-indium-tin alloy in the channel is driven by an electrostatic micropump or thermally induced pressure difference to adjust the conduction path of the microstrip antenna structure and achieve tuning.

[0039] The flow of liquid metal changes the antenna conductor pattern in real time, breaking through the limitations of fixed structures, adapting to multiple frequency bands (such as 5G / 6G switching), and realizing dynamic topology reconfiguration.

[0040] By combining the flexible substrate of PDMS with the high ductility of liquid metal, the liquid metal microfluidic channel is made resistant to bending deformation (>100,000 bends), making it suitable for wearable devices and improving environmental robustness.

[0041] Gallium-indium-tin alloy conductivity >3×10 6 With a S / m ratio and RF loss ≤0.3dB (compared to >1.2dB in traditional switching matrices), low-loss tuning is achieved.

[0042] Micrometer-level channels (100-500μm) enable sub-millimeter-level path adjustment, with frequency tuning resolution up to ±0.5MHz, achieving high-precision control.

[0043] The tuning method of this embodiment, which drives the flow of liquid metal in a liquid metal microfluidic channel by an electrostatic micropump or thermally induced pressure difference, includes the following steps: The external control system generates drive commands to control fluid flow based on frequency response deviation or structural offset information. When an electrostatic micropump is used as the driving method, an asymmetric periodic electric field is applied by the electrodes to generate electrostatic volume force or electroosmotic flow, which causes the liquid metal to flow in a predetermined direction, thereby forming different types of conductive structures. When using the thermally induced pressure difference driving method, a temperature difference is applied at both ends of the channel. A thermal expansion pressure difference is formed by heating one side through a heating device, which guides the liquid metal to migrate in the channel. After the liquid metal flows in the channel, it forms a new conductive shape or an open / closed structure, which adjusts the feed path, electrical length or impedance matching state of the microstrip antenna, thereby achieving dynamic tuning of the input reflection coefficient within the operating frequency band.

[0044] The external control system generates drive commands based on real-time feedback (frequency deviation / structural offset), achieving a closed-loop tuning accuracy of ±0.15dB and realizing intelligent response.

[0045] Electrostatic micropumps: millisecond-level response (<5ms), suitable for frequent frequency hopping scenarios; Thermally induced pressure difference: Passive power consumption (only temperature difference required), suitable for long-term environmental compensation.

[0046] Dual-mode drive adaptability is achieved through electrostatic micropumps and thermally induced pressure differentials.

[0047] Liquid metal flow forms programmable conductive structures (such as branches / rings), supporting more than 10 radiation mode switching and enabling topology free reconfiguration.

[0048] Dynamically adjust the feed path length by ±30% to achieve 1.8-40GHz wideband coverage, optimize return loss by >15dB, and achieve RF performance optimization.

[0049] Implementation Example Electrostatic micropump drive tuning unit Core components: Electrostatic micropump array (silicon-based MEMS pump, driving voltage 0-300V) PDMS microfluidic chip (channel width 300μm, depth 200μm) Gallium-indium-tin alloy (melting point -19℃, viscosity 2×10 at 25℃) -3 Pa·s) Workflow: After the embedded controller detects the need to switch frequency bands, it applies a 250V voltage to the designated micropump. Electrostatic force drives the gallium-indium-tin alloy to flow into the target branch channel, forming a new radiating branch and raising the resonant frequency from 3.5GHz to 5.8GHz. Simultaneously, a piezoelectric sensor provides flow feedback to achieve closed-loop control. The millimeter-wave frequency band switching (28GHz / 39GHz) is completed within 5ms, with an insertion loss of <0.15dB.

[0050] Thermal differential pressure driven tuning unit Core components: Miniature thermoelectric cooler (TEC, temperature difference ±15℃) Dual-channel PDMS flow cavity (main channel 500μm, branch channel 100μm) Temperature sensor (Pt100 thin film) Workflow: When the ambient temperature rise causes antenna frequency deviation, the TEC heats the left cavity (+10℃) and cools the right cavity (-10℃); the liquid metal has a different coefficient of thermal expansion (Ga-In-Sn: 12×10). -6 The resonant frequency is shifted to the right cavity ( / ℃), shortening the effective electrical length and reducing the resonant frequency from 2.6GHz to 2.4GHz, thus compensating for temperature drift. This fully passive temperature drift compensation requires no external power supply and achieves a temperature control accuracy of ±0.5℃.

[0051] The liquid metal microfluidic channel in this embodiment has multiple branch structures inside. Each branch structure is equipped with a controllable microvalve or an electrowetting drive electrode at its end, which is used to selectively open or close the branch channel. The optimization control unit controls the liquid metal fluid to fill the designated branch area according to the frequency response result, thereby realizing the activation or shielding of different antenna structure paths.

[0052] By selectively activating branches, a single antenna can achieve at least four radiation modes (such as omnidirectional / directional / dual-band), replacing traditional antenna arrays and realizing multi-mode radiation switching.

[0053] Microvalves physically isolate unselected channels, with crosstalk between paths ≤-35dB (compared to ≤-25dB for switch matrices), achieving high isolation control.

[0054] The electrowetting electrode adjusts the metal fill length by ±10μm, optimizes return loss by >18dB, and achieves precise impedance matching.

[0055] Shielding redundant paths can suppress harmonic resonances, reduce out-of-band radiation by 8dB, and improve anti-interference capabilities.

[0056] Implementation Example Module 1: Electrowetting Electrode Branch Control System (Millimeter Wave Band) Core components: Indium tin oxide (ITO) electrowetting electrode array (20 μm linewidth) High voltage multiplexer (ADG5421F, withstand voltage 300V) PDMS branched microchannels (main channel 400μm, branches 100μm×4) Workflow: When detecting the 39GHz band demand, the control unit applies a 150V DC voltage to the ITO electrode of the target branch; the electrowetting effect reduces the surface tension of the Ga-In-Sn alloy, causing it to fill the λ / 4 resonant branch; the millimeter-wave radiation structure is activated, while other branches are shielded, with a center frequency accuracy of ±0.3GHz. Switching time: 8ms; power consumption: 0.1μJ / switch (electrode capacitance 5pF).

[0057] Module 2: Shape Memory Alloy (SMA) Microvalve Control System (Vibration-Resistant Scenarios) Core components: Nickel-Titanium SMA microvalve (300μm diameter, phase transition temperature 70℃) Miniature heating resistor (platinum wire, 50Ω resistance) Strain sensor (piezoelectric thin film PVDF) Workflow: When the vehicle-mounted antenna experiences frequency deviation due to vibration, the strain sensor triggers the heating resistor of the corresponding branch; the SMA microvalve contracts upon heating to open the channel, and liquid metal fills the deformation area of ​​the compensation structure; simultaneously, irrelevant branches are closed to suppress mode distortion and maintain pattern gain >8dBi. Vibration compensation range: 10-500Hz; operating temperature: -40℃~125℃.

[0058] Discrete reconstruction of the antenna topology is achieved through the selective switching of branch channels. Electrowetting electrodes provide microsecond-level precise filling, while SMA microvalves ensure reliability in extreme environments. Together, they address the anti-interference and multi-mode switching requirements in mobile scenarios such as 5G vehicles and drones. The ITO electrode array and SMA actuator module can be directly integrated into microwave circuit boards, suitable for base station beamforming and satellite multi-beam antennas.

[0059] In this embodiment, resistance sensing electrodes or optical reflection detection structures are set at key node positions of the liquid metal microfluidic channel to monitor the liquid metal filling status in the channel in real time. The detection results are uploaded to the optimization control unit and compared with the expected conduction path to determine whether the target path formation has been completed.

[0060] Real-time monitoring of key nodes (branch intersections / path ends) with a filling and positioning accuracy of ±0.15mm avoids path defects and achieves zero blind spots in the reconstruction process.

[0061] By comparing the expected path with the actual filling state, the driving parameters (voltage / temperature difference) are dynamically adjusted to compensate for manufacturing tolerances and enhance fault-tolerant control.

[0062] Resistance sensing resists light / magnetic interference, optical detection penetrates liquid metal microfluidic channels, adapts to operating conditions from -40℃ to 85℃, and achieves resistance to environmental interference.

[0063] The resistive sensing electrode or optical reflection detection structure can complete path verification within 5ms, which is 100 times faster than traditional VSWR detection. It supports millisecond-level frequency hopping, which improves the response speed.

[0064] Implementation Example Module 1: Four-wire resistance sensing system (high-precision scenario) Core components: Platinum microelectrode pairs (20 μm linewidth, 50 μm spacing) Constant current source + lock-in amplifier (AD5933 impedance chip) PDMS microchannel node (covering electrode area) Workflow: A constant current source outputs 1mA to the electrode, and a lock-in amplifier measures the node resistance in real time after liquid metal filling; when the detected resistance drops from ∞ to 8mΩ (gallium-indium-tin alloy resistivity 2.9×10⁻⁶), the resistance is measured. -7 Ω·m), determine path continuity; if the deviation from the expected value is >10%, trigger the micro-pump to supplement flow until the error is <±2%. Resolution: 0.1mΩ (equivalent to 0.05mm filling length); response time: 1ms.

[0065] Module 2: Optical Reflection Detection System (Applicable to Transparent Structures) Core components: Vertical-cavity surface-emitting laser (VCSEL, wavelength 850nm) Four-quadrant photodiode (OPT101) PDMS microchannels (1mm thickness, >92% light transmittance) Workflow: The VCSEL emits a laser that penetrates PDMS to irradiate the liquid metal surface, and a photodiode receives the reflected light intensity. The reflectivity is 4% (PDMS-air interface) when unfilled, and increases to 89% (PDMS-metal interface) after filling. The position of the metal front is calculated based on the laser spot offset, with an accuracy of ±0.1mm, and fed back to the control unit in real time to generate a path topology map. Sampling rate: 10kHz; Power consumption: <15mW.

[0066] In this embodiment, the AI ​​image recognition module acquires images of the antenna pattern area at a high frame rate during the liquid metal microfluidic filling process. It then analyzes the formation of the conduction path by comparing the image features with the initial reference image. The analysis results, together with the frequency response data provided by the near-field electromagnetic detection module, serve as the input to the optimization control unit, thereby achieving coordinated tuning of visual and electromagnetic feedback.

[0067] By deeply integrating AI visual monitoring with near-field electromagnetic detection, the core pain points of "uncontrollable shape and difficult performance verification" in the reconfiguration process of liquid metal antennas are overcome. Visual feedback ensures accurate reproduction of geometric paths, while electromagnetic feedback verifies that the radio frequency functions meet the standards. The closed-loop collaboration of the two enables reliable tuning under complex operating conditions, providing a foundation for the industrial application of 5G / 6G reconfigurable antennas.

[0068] The optimization control unit in this embodiment includes a deep learning model that integrates a multilayer perceptron network and a convolutional neural network, used to implement the following real-time dynamic tuning steps: A. The structural parameter extraction module extracts the wiring shape, feed line width, and substrate integrated waveguide via layout of the microstrip antenna to generate structural feature data. B. The input reflection coefficient (S) is collected through the near-field electromagnetic detection module. 11 ), phase response and frequency shift, to generate electrical performance feedback data; C. The AI ​​image recognition module acquires antenna images and compares them with the design pattern to generate image deviation data; D. Input the above three types of data into the deep learning model. The model consists of a multilayer perceptron subnetwork and a convolutional neural subnetwork, which are used to model electromagnetic response behavior and identify structural anomalies, respectively. E. The optimization control unit outputs at least one of the following tuning control signals: E1, the driving signal that controls the morphology of liquid metal in the liquid metal microfluidic channel; E2, activates the electrical, optical, and thermal excitation signals of the tuning material; E3, a modulation signal acting on a piezoelectric or thermoelectric thin film structure; F. The tuning control signal is used to drive the coordinated adjustment of the antenna structure and tuning materials to continuously maintain the input reflection coefficient (S) under conditions of structural displacement, electrical performance drift, or environmental disturbance. 11The engineering matching standard is below -10dB, thereby ensuring stable radiation performance in the target frequency band.

[0069] A deep learning-driven multi-source sensing-cooperative execution architecture overcomes the industry challenge of antenna performance instability under complex operating conditions. A dual-network architecture combining a multilayer perceptron network and a convolutional neural network enables joint deduction of electromagnetic response and physical deformation. Three actuators (liquid metal / smart material / thin film) perform closed-loop corrections for the conduction path, dielectric parameters, and mechanical structure, ultimately ensuring a robust engineering performance of S11 ≤ -10dB across the entire frequency band under extreme disturbances. This provides a reliable technical foundation for scenarios such as 5G / 6G base stations and satellite payloads.

[0070] Preferably, the AI ​​image recognition module includes an embedded image acquisition device and a lightweight convolutional neural network model. The module identifies linewidth deviations, contour defects, and cracks in microstrip antenna patterns through edge detection and shape matching algorithms, with an accuracy better than 20 micrometers. By using AI vision as a "micro-scale microscope," it captures physical defects (cracks / deformations) with an accuracy of 20μm, preventing failure at its source.

[0071] The near-field electromagnetic detection module includes multiple near-field coupling probes, which are arranged at intervals of less than 5 mm on the edge region of the microstrip antenna structure to collect the input reflection coefficient (S). 11 Phase response and frequency drift data, with a sampling period of less than 100 milliseconds.

[0072] The near-field probe acts as an "electromagnetic stethoscope," quantifying S11 / phase / frequency offset in real time through 100ms-level full-band scanning. Both provide high-precision, low-latency, and multi-dimensional input data for the deep learning optimization control unit, ultimately ensuring a hard target of S11≤-10dB under complex disturbances, meeting the high-reliability requirements of 5G base stations / satellite communications and other scenarios.

[0073] On the other hand, a millimeter-wave FPC multi-domain adaptive tuning antenna is also provided, which is used to implement the above-mentioned tuning method. The antenna includes: The flexible microstrip antenna structure features a biomimetic feed pattern optimized through topology self-growth evolution to meet the radiation characteristics of the target frequency band. Topology evolution is performed based on power density distribution, and a biomimetic growth method is used to optimize the current path, improving radiation efficiency, enhancing broadband and multi-band coverage, and adapting to structural disturbances. The flexible microstrip antenna structure is a polyimide copper-clad laminate, such as the DuPont PyraluxAP. The biomimetic pattern is generated collaboratively using Ansys HFSS electromagnetic simulation and the MATLAB optimization toolbox.

[0074] Tuning material units are set in the antenna structure. The tuning material is responsive to external physical stimuli to adjust local structural or electrical performance parameters. The tuning material produces changes in dielectric constant or conductivity in response to external stimuli (such as temperature, humidity, stress, electric field, etc.) to achieve passive or semi-active local tuning, thereby enhancing environmental adaptability and structural self-healing ability.

[0075] The tuning material unit includes at least one of the following materials: photoresponsive materials, such as polyethylene terephthalate liquid crystal and photopolymer composite film; thermally responsive materials, such as vanadium oxide (VO2) thin film; electrically responsive materials, such as fluorine-doped graphene composite electrostrictive polymer; and phase change materials, such as germanium antimony tellurium (Ge2Sb2Te5) composite material.

[0076] The liquid metal microfluidic tuning unit includes at least one closed channel made of flexible material and liquid metal filled in the closed channel. The channel is deployed in a critical area of ​​the antenna and can achieve controllable migration of the metal fluid under electrostatic or thermal actuation to dynamically adjust the conduction path. By controlling the migration of the liquid metal using the flexible channel and adjusting the conduction state of the path, dynamic frequency tuning is achieved. It also allows for precise control of the feed path and matching impedance, enabling rapid response to changing operating conditions.

[0077] The closed channel is a sealed structure made of flexible polydimethylsiloxane (PDMS), with a channel cross-sectional width of 100 micrometers to 500 micrometers. The liquid metal microfluidic channel is filled with a gallium-indium-tin alloy, which is driven to flow in the channel by an electrostatic micropump or thermal pressure difference.

[0078] The AI ​​image recognition module and the near-field electromagnetic detection module are used to acquire antenna geometric deviation images and spectral response data, respectively. Image recognition identifies antenna deformation or manufacturing deviations, revealing manufacturing errors or geometric deviations during use, enabling intelligent compensation. The near-field distribution and spectral response of the antenna provide real-time performance evaluation, offering information such as frequency offset and VSWR for feedback control, thus enhancing the system's intelligent response capabilities.

[0079] The AI ​​image recognition module combines an embedded camera (OV5640) and an NVIDIA Jetson Nano, an embedded artificial intelligence computing platform from NVIDIA. The algorithms utilize a high-efficiency defect recognition engine based on the MobileNetV3 lightweight model and a sub-pixel contour analysis tool based on OpenCV Sobel edge detection. The near-field electromagnetic detection module uses a PWR-6G magnetic coupling probe.

[0080] The optimization control unit is used to fuse and analyze image recognition results and near-field response parameters, and outputs tuning control signals based on the tuning database to achieve adaptive optimization of the antenna structure and frequency matching tuning. By fusing image and detection data and combining them with the tuning database to generate control strategy signals, adaptive, multi-domain collaborative tuning is achieved to maintain optimal communication performance.

[0081] The optimized control unit is an NXPi.MX8MPlus processor, with an architecture that includes a multilayer perceptron subnetwork (Scikit-learn) to process electromagnetic data and a convolutional neural subnetwork (PyTorchMobile) to analyze images.

[0082] In use, when environmental disturbances or frequency band switching cause antenna performance degradation, the AI ​​image recognition module collects images of geometric deviations on the antenna surface in real time, while the near-field electromagnetic detection module simultaneously acquires spectral response data. The optimization control unit fuses the two types of data to generate tuning commands, driving a triple tuning mechanism to respond collaboratively—the tuning material unit autonomously adjusts its local electrical parameters and structural morphology under physical stimulation, and the liquid metal microfluidic tuning unit migrates and reconstructs the conduction path under electrostatic / thermal driving. Finally, frequency matching is achieved through the biomimetic topology of the flexible microstrip antenna structure. The entire process is completed in a closed loop within 20ms, achieving adaptive optimization and maintaining stable radiation characteristics with an input reflection coefficient ≤-10dB in the 24.25-43.5GHz frequency band.

[0083] The above description is only a preferred embodiment of the present invention. For those skilled in the art, there will be changes in the specific implementation and application scope based on the ideas of the present invention. The content of this specification should not be construed as a limitation of the present invention.

Claims

1. A tuning method for a millimeter-wave FPC multi-domain adaptive tuning antenna, characterized in that, Includes the following steps: S1. Construct the initial topology and frequency target model of the microstrip antenna structure; S2. Using the topology self-growth algorithm, the structure evolves based on the power density distribution to generate an antenna pattern with a biomimetic shape. S3. Embed a tuning material with physical response characteristics in the microstrip antenna structure. Under the action of external physical stimuli, the tuning material itself will change its structure or conductivity to adjust the local matching state, input impedance or resonant frequency of the antenna. S4. Liquid metal microfluidic channels are set in key areas of the microstrip antenna structure. The channels are filled with liquid metal, and the distribution position and shape of the liquid metal in the channels are controlled to change the equivalent conduction path of the microstrip antenna. S5. An AI image recognition module and a near-field electromagnetic detection module are set on the surface of the microstrip antenna. The AI ​​image recognition module is used to identify the geometric deviation image of the microstrip antenna, and the near-field electromagnetic detection module is used to collect spectral response characteristic parameters. S6. The image recognition results and near-field response parameters are jointly input to the optimization control unit, and dynamic parameter matching and tuning command output are executed based on the historical tuning database to complete the automatic structural optimization and matching tuning of the microstrip antenna in the millimeter-wave band.

2. The tuning method according to claim 1, characterized in that, The self-growing topology algorithm establishes a finite element analysis model based on power density distribution, uses Bezier curves or spline functions to describe the shape of the antenna boundary, and performs thickening, branching or shrinking operations on the edge contour of the microstrip antenna structure in each iteration, thereby forming a biomimetic leaf vein-like topology and realizing the generation of distributed feed paths.

3. The tuning method according to claim 1, characterized in that, The tuning material with physical response characteristics induces changes in conductivity, dielectric constant, structural size deformation, or phase transition behavior through external physical stimuli, thereby adjusting the feed path, electrical length, input impedance, resonant frequency, or return loss of the microstrip antenna structure, and possesses dual self-tuning capabilities in terms of electrical performance and structural morphology.

4. The tuning method according to claim 1, characterized in that, The liquid metal microfluidic channel is a sealed structure made of flexible polydimethylsiloxane with a cross-sectional width of 100 micrometers to 500 micrometers. The liquid metal inside the liquid metal microfluidic channel is a gallium-indium-tin alloy. The flow of the gallium-indium-tin alloy in the channel is driven by an electrostatic micropump or thermally induced pressure difference to adjust the conduction path of the microstrip antenna structure and achieve tuning.

5. The tuning method according to claim 4, characterized in that, The tuning method for driving the flow of liquid metal in a liquid metal microfluidic channel by an electrostatic micropump or thermally induced pressure difference includes the following steps: The external control system generates drive commands to control fluid flow based on frequency response deviation or structural offset information. When an electrostatic micropump is used as the driving method, an asymmetric periodic electric field is applied by the electrodes to generate electrostatic volume force or electroosmotic flow, which causes the liquid metal to flow in a predetermined direction, thereby forming different types of conductive structures. When using the thermally induced pressure difference driving method, a temperature difference is applied at both ends of the channel. A thermal expansion pressure difference is formed by heating one side through a heating device, which guides the liquid metal to migrate in the channel. The liquid metal flows in the channel and forms a new conductive shape or open / closed structure, which adjusts the feed path, electrical length or impedance matching state of the microstrip antenna, thereby achieving dynamic tuning of the input reflection coefficient within the operating frequency band.

6. The tuning method according to claim 4, characterized in that, The liquid metal microfluidic channel has multiple branch structures inside, and each branch structure is equipped with a controllable microvalve or electrowetting drive electrode at the end to selectively open or close the branch channel. The optimization control unit controls the liquid metal fluid to fill the designated branch area according to the frequency response result, thereby realizing the activation or shielding of different antenna structure paths.

7. The tuning method according to claim 4, characterized in that, Resistance sensing electrodes or optical reflection detection structures are set at key node positions in the liquid metal microfluidic channel to monitor the liquid metal filling status in the channel in real time. The detection results are uploaded to the optimization control unit and compared with the expected conduction path to determine whether the target path has been formed.

8. The tuning method according to claim 4, characterized in that, The AI ​​image recognition module acquires images of the antenna pattern area at a high frame rate during the liquid metal microfluidic filling process, and analyzes the formation of the conduction path by comparing the image features with the initial reference image. The analysis results, together with the frequency response data provided by the near-field electromagnetic detection module, serve as the input to the optimization control unit, thereby achieving coordinated tuning of visual feedback and electromagnetic feedback.

9. The tuning method according to claim 1, characterized in that, The optimization control unit includes a deep learning model that integrates a multilayer perceptron network and a convolutional neural network, used to implement the following real-time dynamic tuning steps: A. The structural parameter extraction module extracts the wiring shape, feed line width, and substrate integrated waveguide via layout of the microstrip antenna to generate structural feature data. B. The input reflection coefficient, phase response and frequency shift are collected by the near-field electromagnetic detection module to generate electrical performance feedback data; C. The AI ​​image recognition module acquires antenna images and compares them with the design pattern to generate image deviation data; D. Input the above three types of data into a deep learning model, which consists of a multilayer perceptron subnetwork and a convolutional neural subnetwork, used to model electromagnetic response behavior and identify structural anomalies, respectively. E. The optimization control unit outputs at least one of the following tuning control signals: E1, the driving signal that controls the morphology of liquid metal in the liquid metal microfluidic channel; E2, activates the electrical, optical, and thermal excitation signals of the tuning material; E3, a modulation signal acting on a piezoelectric or thermoelectric thin film structure; F. The tuning control signal is used to drive the antenna structure and tuning material to adjust in coordination, so as to continuously maintain the engineering matching standard of input reflection coefficient below -10dB under the conditions of structural offset, electrical performance drift or environmental disturbance, thereby ensuring the stable radiation performance of the target frequency band.

10. A millimeter-wave FPC multi-domain adaptive tuning antenna, characterized in that, The antenna is used to implement the tuning method as described in any one of claims 1 to 9, the antenna comprising: The flexible microstrip antenna structure features a biomimetic feed pattern optimized through topological self-growth evolution to meet the radiation characteristics of the target frequency band. A tuning material unit is disposed in the antenna structure. The tuning material is responsive to external physical stimuli in order to adjust local structural or electrical performance parameters. A liquid metal microfluidic tuning unit includes at least one closed channel made of a flexible material and liquid metal filled in the closed channel. The channel is located in a critical area of ​​the antenna and can achieve controllable migration of the metal fluid under electrostatic or thermal actuation to dynamically adjust the conduction path. The AI ​​image recognition module and the near-field electromagnetic detection module are used to collect antenna geometric deviation images and spectral response data, respectively. The optimization control unit is used to fuse and analyze the image recognition results and near-field response parameters, and output tuning control signals based on the tuning database to complete the adaptive optimization of the antenna structure and frequency matching tuning.

Images