Variable-focus optical lens module and imaging equipment thereof

Abstract

The invention discloses a variable-focus optical lens module and imaging equipment thereof. The variable-focus optical lens module comprises a module shell, a sub-lens group, a magnetic suspension multi-degree-of-freedom driving mechanism, an AI cooperative control module, an environment sensing module and a personalized storage module. The inner wall of the module shell is provided with an electromagnetic shielding layer, flexible shading fluff and a miniature heat dissipation channel, and a graphene heat dissipation film is attached to the outer side of the module shell. The sub-lens group is composed of a lens fixing seat, an optical lens, a permanent magnet ring and a fiber grating sensor. The magnetic suspension multi-degree-of-freedom driving mechanism realizes suspension, movement and stable control of the sub-lens group in a three-dimensional space through a Helmholtz coil, an eddy current damper and a Hall sensor. And the AI cooperative control module comprises scene perception, AI decision and a dynamic compensation algorithm, dynamically outputs driving parameters in combination with multi-source data, and realizes self-adaptive optimization of an imaging effect. The environment sensing module collects temperature, humidity and vibration data, and the personalized storage module stores user parameters and an optimization model to form complete closed-loop control.

Description

Technical Field

[0001] This invention relates to the field of motorized zoom lens technology, and more particularly to a zoomable optical lens module and its imaging device. Background Technology

[0002] With the widespread adoption of mobile terminals, wearable devices, and intelligent imaging systems, the performance of optical lens modules, as core imaging components, directly determines image quality and user experience. In diverse application scenarios, optical lens modules not only need to possess high-resolution imaging capabilities but also need to meet the design requirements of miniaturization, lightweight design, and low power consumption.

[0003] In recent years, users have increasingly higher demands for the shooting experience of smart devices, especially in areas such as zoom imaging, adaptability to complex lighting environments, and fast focusing stability. However, in achieving zoom capabilities, optical lens modules often face the following challenges: 1. Conflict between structural integration and compactness: Within a limited module space, multiple sets of lenses and their driving mechanisms need to be accommodated, which can easily lead to complex structural stacking and prominent issues of heat dissipation and electromagnetic compatibility.

[0004] 2. High difficulty in precision drive and stable control: The movement of the lens assembly usually requires a high-precision drive and feedback mechanism to ensure image sharpness and smoothness during zooming. However, in multi-degree-of-freedom adjustment, vibration, offset, or response delay are often present.

[0005] 3. Insufficient environmental adaptability: In complex application scenarios, environmental factors such as temperature, humidity and external vibration can affect the performance of lens materials and drive units, resulting in fluctuations in image quality.

[0006] 4. Limited level of intelligence: Traditional optical modules mostly rely on fixed algorithms or single control logic, lacking the ability to adaptively adjust based on real-time scene characteristics, making it difficult to meet the imaging needs of multiple scenarios.

[0007] Therefore, how to achieve high-precision multi-degree-of-freedom lens adjustment in a compact module structure, and improve the stability and intelligence of the module in complex environments, has become a key issue that urgently needs to be addressed in the field of optical lens modules. Summary of the Invention

[0008] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a zoomable optical lens module and its imaging device, which can realize high-precision, stable and intelligent zoom adjustment of the optical lens, and significantly improve imaging quality and environmental adaptability.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a variable zoom optical lens module, comprising: Module casing; N camera groups are coaxially arranged inside the module housing, where N is an integer greater than or equal to 2; each camera group includes a lens holder, M optical lenses fixed inside the lens holder, a permanent magnet ring sleeved on the outside of the lens holder, and a miniature fiber optic sensor embedded inside the lens holder, where M is an integer greater than or equal to 3. The magnetic levitation multi-degree-of-freedom drive mechanism, the number of which corresponds to the number of the split-screen group, includes three pairs of Helmholtz coils fixed to the inner wall of the module housing, namely the X-axis, Y-axis and Z-axis, an eddy current damper disposed inside the Helmholtz coils, and a miniature Hall sensor disposed between the Helmholtz coils and the split-screen group. The AI ​​collaborative control module includes a hardware layer and an algorithm layer. The hardware layer includes a main MCU, an AI acceleration chip, and an image signal processor (ISP). The algorithm layer includes a scene perception algorithm, an AI decision-making algorithm, and a dynamic compensation algorithm. The environmental sensing module includes a triaxial accelerometer and a temperature and humidity sensor; A personalized storage module is used to store user-defined parameters and scene feature-driving parameter mapping models; The AI ​​collaborative control module is electrically connected to the magnetic levitation multi-degree-of-freedom drive mechanism, the environmental perception module, and the personalized storage module, forming a collaborative control closed loop. It is used to control the Helmholtz coil and eddy current damper based on sensor data to drive the split-camera group to perform multi-axis fine-tuning.

[0010] Furthermore, the magnetic levitation multi-degree-of-freedom drive mechanism is configured such that the suspension gap of the split-camera group is maintained between 0.1mm and 0.3mm during the drive process, and the control accuracy of the suspension gap is ±0.01mm; The damping coefficient of the eddy current damper can be dynamically adjusted by the AI ​​collaborative control module within the range of 0.01 N·s / m to 0.05 N·s / m.

[0011] Furthermore, the sampling frequency of both the miniature Hall sensor and the miniature fiber Bragg grating sensor is not less than 1 kHz; The sampling frequency of the triaxial accelerometer is not less than 100Hz.

[0012] Furthermore, the pre-trained scene feature-driving parameter mapping model is trained on a training dataset containing more than 100,000 samples; The variables covered by the training dataset include: The brightness range is 0 to 255 cd / m², the object movement speed range is 0 to 5 m / s, the shooting distance range is 0.1 m to 10 m, and different ambient temperature and vibration conditions are also considered.

[0013] Furthermore, the dynamic compensation algorithm is configured to be based on the temperature change monitored by the miniature fiber Bragg grating sensor. Calculate a compensation parameter And based on the compensation parameters Perform at least one of the following compensation operations: The current of the Helmholtz coil is compensated, or the spacing of the shot group is compensated; wherein, the compensation parameter... Defined by the following formula: ; Where A is the compensation amplitude scaling factor, B is the error function scaling factor, C is the Bessel function scaling factor, and D is... Function parameter scaling factor, E is logarithmic The weighting coefficients of the product, F is the scaling factor for the logarithmic function parameters, and G is... The scaling factor for the function parameters, H, is Function parameter offset, The change in temperature Let be the error function. It is a first-order Bessel function. for function, It is the natural logarithm function. for Function; and, the current compensation ratio for the Helmholtz coil. Compensation amount for the spacing of the storyboard group ,in and As a preset constant, The value ranges from 0.05 to 0.10. The value ranges from 0.02 to 0.05.

[0014] Furthermore, the personalized storage module is a NOR flash memory; The AI ​​decision-making algorithm supports incremental learning, which is used to iteratively optimize the scene feature-driving parameter mapping model based on the historical usage data, and improve the model accuracy by 0.5% to 1% after each optimization.

[0015] Furthermore, the collaborative control closed loop is configured to operate according to the following process: During the sensing phase, data from image sensors, miniature Hall sensors, miniature fiber Bragg grating sensors, temperature and humidity sensors, and triaxial accelerometers are collected simultaneously. During the decision-making phase, the collected data is processed through AI decision-making algorithms, and driving parameters and compensation parameters are output. During the execution phase, the main MCU controls the Helmholtz coil and eddy current damper according to the driving parameters and compensation parameters to drive the split-camera group to perform multi-axis fine-tuning. During the feedback phase, the status data and imaging effect data of the lens group after adjustment are obtained through miniature Hall sensors and image sensors; During the optimization phase, the AI ​​decision algorithm compares the target imaging effect with the actual imaging effect. If the deviation exceeds 5%, the driving parameters are optimized a second time and the process is re-executed until the deviation does not exceed 5%. The total response time of the entire process from the first execution to finally meeting the deviation requirement is controlled within 50ms, and the number of iterations of the second optimization does not exceed 3.

[0016] Furthermore, the module shell is made of carbon fiber composite material or high-strength aluminum alloy, and its wall thickness is 0.5mm to 1.2mm; The electromagnetic shielding layer is a copper foil layer plated on the inner wall of the module housing, with a thickness of 0.02 mm to 0.05 mm.

[0017] An imaging device includes a housing, an image sensor, and a processing unit, wherein the imaging device integrates a variable-focus optical lens module as described above.

[0018] The beneficial effects of this invention are: This invention effectively improves the module's anti-interference capability, imaging clarity, and heat dissipation performance by incorporating an electromagnetic shielding layer, light-blocking floss, and heat dissipation structure within the module's housing, thereby ensuring long-term stable operation. The introduction of a fiber optic grating sensor into the lens assembly structure enables real-time temperature monitoring and compensation, reducing the impact of thermal effects on imaging. This invention employs a magnetic levitation multi-degree-of-freedom drive mechanism, enabling the lens assembly to achieve high-precision levitation and movement in three-dimensional space, avoiding traditional mechanical friction and wear. At the same time, in conjunction with an eddy current damper and a Hall sensor, it improves the stability and accuracy of the lens adjustment process. The AI ​​collaborative control module achieves adaptive optimization of the driving process through scene perception, intelligent decision-making, and dynamic compensation, ensuring ideal imaging results in different scenarios. The personalized storage function supports model self-learning and user parameter memorization, enhancing the system's flexibility and adaptability. The invention adopts a closed-loop control method of perception-decision-execution-feedback-optimization, which enables the imaging process to be adjusted and continuously optimized in real time, effectively improving the imaging quality, stability and intelligence level of the zoom optical lens module. Attached Figure Description

[0019] Figure 1 This is a control structure diagram of the variable zoom optical lens module in this invention.

[0020] Reference numerals: 1. Split-lens group; 11. Optical lens; 12. Permanent magnet ring; 13. Miniature fiber optic grating sensor; 2. Magnetic levitation multi-degree-of-freedom drive mechanism; 21. Helmholtz coil; 22. Eddy current damper; 23. Miniature Hall sensor; 3. AI collaborative control module; 31. Hardware layer; 32. Algorithm layer; 4. Environmental perception module; 5. Personalized storage module. Detailed Implementation

[0021] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Identical components are denoted by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom surface," "top surface," "inner," and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0022] Example 1, referring to Figure 1 This is the first embodiment of the present invention. This embodiment provides a variable zoom optical lens module, including a module housing, a split lens group 1, a magnetic levitation multi-degree-of-freedom drive mechanism 2, an AI collaborative control module 3, an environmental perception module 4, and a personalized storage module 5.

[0023] The module shell is made of aluminum-magnesium alloy, with an electromagnetic shielding layer and flexible light-blocking floss arranged sequentially on its inner wall, forming a micro heat dissipation channel inside. A graphene heat dissipation film is attached to the outside to achieve anti-interference, suppression of stray light and efficient heat dissipation.

[0024] The lens assembly 1 is arranged coaxially along the optical axis and consists of three groups. Each group comprises a lens holder, five optical lenses 11, a permanent magnet ring 12, and an embedded miniature fiber Bragg grating sensor 13. The optical lenses 11 are made of low-dispersion glass to ensure image sharpness; the permanent magnet ring 12 is made of neodymium iron boron material to provide a stable magnetic field; and the fiber Bragg grating sensor monitors the lens temperature in real time for temperature compensation.

[0025] The magnetic levitation multi-degree-of-freedom drive mechanism 2 corresponds to each shot group 1 and includes three pairs of Helmholtz coils 21, eddy current dampers 22, and miniature Hall sensors 23. The Helmholtz coils 21 interact with the permanent magnet ring 12 to enable the shot group 1 to levitate and move in three-dimensional space; the eddy current dampers 22 are used to suppress vibrations during high-speed movement; and the Hall sensors collect the motion state data of the shot group 1 in real time.

[0026] The AI ​​collaborative control module 3 includes a main MCU, an AI acceleration chip, and an image signal processor (ISP). It runs scene perception algorithms, AI decision-making algorithms, and dynamic compensation algorithms, respectively responsible for data fusion, drive parameter generation, and temperature compensation. The environmental perception module 4 includes a triaxial accelerometer and a temperature and humidity sensor to monitor the vibration and temperature and humidity of the environment in which the module is located. The personalized storage module 5 stores user parameters, optimization models, and historical data.

[0027] This module can be applied to terminal devices such as smartphones and drone imaging systems.

[0028] Working principle of Example 1: During the sensing phase, the image signal processor, Hall sensor, fiber optic grating sensor, temperature and humidity sensor, and accelerometer simultaneously acquire data. During the decision-making stage, the AI ​​decision-making algorithm integrates image features, driving state data and environmental parameters to generate driving parameters such as the current of the Helmholtz coil 21, the damping strength of the eddy current damper 22 and the moving step size of the split-screen group 1. During the execution phase, the main MCU controls the Helmholtz coil 21 and the eddy current damper 22 to enable the shot group 1 to perform multi-degree-of-freedom adjustment. During the feedback phase, the Hall sensor and ISP continuously acquire position and imaging data of the first shot group. During the optimization phase, the AI ​​algorithm compares the target imaging effect with the actual imaging result. If the deviation exceeds 5%, the driving parameters are updated and readjusted until the deviation meets the requirements.

[0029] Technical effects of Example 1: The module housing features electromagnetic shielding, light-blocking, and heat dissipation designs, resulting in clearer imaging and ensuring stable operation under high loads. The first lens group integrates a fiber optic grating sensor, which can realize real-time temperature compensation and reduce the impact of thermal effects on image quality. Magnetic levitation drive enables non-contact levitation and high-precision movement in three-dimensional space, avoiding mechanical friction and wear. At the same time, the eddy current damper 22 and Hall sensor ensure the stability and accuracy of the adjustment. AI Collaborative Control Module 3 achieves adaptive adjustment of imaging effects through a closed-loop process of perception, decision-making, execution, feedback, and optimization, and automatically optimizes when the deviation exceeds the threshold to ensure imaging accuracy. The personalized storage module 5 supports self-learning and parameter memorization, enabling the module to adapt to user habits and different application environments, further improving its intelligence level.

[0030] Example 2 is the second embodiment of the present invention, which provides a variable zoom optical lens module further optimized based on Example 1. The module includes a module housing, a split lens group 1, a magnetic levitation multi-degree-of-freedom drive mechanism 2, an AI collaborative control module 3, an environmental perception module 4, and a personalized storage module 5.

[0031] The module shell is made of carbon fiber composite material or high-strength aluminum alloy, with a wall thickness controlled between 0.5mm and 1.2mm, ensuring both overall strength and lightweight design. The inner wall of the module shell is coated with a copper foil layer with a thickness of 0.02mm to 0.05mm as an electromagnetic shielding layer, effectively suppressing external electromagnetic interference. The module shell incorporates flexible light-shielding fibers and micro-heat dissipation channels, while a graphene heat dissipation film is attached to the outer side, achieving comprehensive optimization of optical, heat dissipation, and electromagnetic compatibility performance.

[0032] The split-lens group 1 is the same as in embodiment 1, and is arranged coaxially along the optical axis. Each group includes a lens holder, an optical lens 11, a permanent magnet ring 12, and a fiber optic grating sensor. The fiber optic grating sensor has a sampling frequency of not less than 1 kHz, which enables real-time high-frequency monitoring of temperature changes.

[0033] The magnetic levitation multi-degree-of-freedom drive mechanism 2 further specifies that during the driving of the shot assembly 1, the suspension gap is maintained between 0.1 mm and 0.3 mm, with a control accuracy of ±0.01 mm. This mechanism includes three pairs of Helmholtz coils 21, eddy current dampers 22, and Hall effect sensors. The damping coefficient of the eddy current dampers 22 can be dynamically adjusted by the AI ​​collaborative control module 3 according to imaging requirements and environmental conditions, within the range of 0.01 N·s / m to 0.05 N·s / m. The Hall effect sensor has a sampling frequency of no less than 1 kHz and is used to monitor the motion state and suspension gap of the shot assembly 1 in real time.

[0034] The environmental perception module 4 includes a triaxial accelerometer with a sampling frequency of no less than 100Hz, capable of capturing environmental vibrations in real time. The AI ​​collaborative control module 3 and the personalized storage module 5 are the same as in Example 1.

[0035] In addition, this embodiment also provides an imaging device, which includes a housing, an image sensor and a processing unit, wherein the aforementioned variable zoom optical lens module is integrated, enabling the imaging device to have high-precision intelligent zoom capability.

[0036] Working principle of Example 2: The working principle of this embodiment is similar to that of Embodiment 1, mainly including five stages: perception, decision-making, execution, feedback, and optimization. The difference lies in: During the sensing phase, Hall sensors and fiber optic grating sensors collect position and temperature information of the first shot group in real time at a sampling frequency of no less than 1 kHz, while a triaxial accelerometer monitors environmental vibrations at a frequency of no less than 100 Hz. During the decision-making stage, the AI ​​decision-making algorithm, based on the integration of the above high-frequency data, performs precise calculations on the current control of the Helmholtz coil 21 and the damping coefficient of the eddy current damper 22 to maintain the suspension gap of the shot group from 10.1mm to 0.3mm and ensure a control accuracy of ±0.01mm. During the execution phase, the main MCU controls the Helmholtz coil 21 and the eddy current damper 22 according to the AI ​​algorithm output parameters to achieve stable suspension and precise adjustment of the shot group 1. During the feedback and optimization phase, the system compares the actual imaging effect with the target imaging effect. When the deviation exceeds the set threshold, the AI ​​algorithm will dynamically adjust the current and damping parameters until the deviation meets the requirements.

[0037] Technical effects of Example 2: The module shell is made of carbon fiber composite material or high-strength aluminum alloy, which balances lightweight and high strength. Combined with copper foil shielding layer and graphene heat dissipation film, the module still has excellent electromagnetic shielding and heat dissipation capabilities in complex environments.

[0038] The suspension gap is precisely controlled between 0.1mm and 0.3mm, with an accuracy of ±0.01mm, ensuring the stable suspension and high-precision movement of the first lens group in three-dimensional space, thereby improving imaging stability.

[0039] The damping coefficient of the eddy current damper 22 can be dynamically adjusted by AI, so that the response speed and stability of the lens can be balanced in different scenarios, avoiding the impact of vibration or shaking on imaging.

[0040] High-frequency sampling by Hall effect sensors and fiber Bragg grating sensors improves the real-time performance and accuracy of feedback control, ensuring more efficient dynamic compensation.

[0041] The triaxial accelerometer samples environmental vibrations at high frequency, enabling the module to quickly and adaptively adjust to external disturbances, further enhancing its anti-interference performance.

[0042] When integrated into imaging devices, it enables high-precision, low-latency intelligent zoom in smartphones, drones, or security devices, significantly improving image quality and user experience.

[0043] Example 3, the third embodiment of the present invention, provides a further optimized variable-focus optical lens module compared to Example 3. This module includes a module housing, a lens assembly 1, a magnetic levitation multi-degree-of-freedom drive mechanism 2, an AI collaborative control module 3, an environmental perception module 4, and a personalized storage module 5. The specific structure is the same as in Examples 1 and 2, and will not be described again. This embodiment focuses on improvements to the AI ​​model, dynamic compensation algorithm, and storage and optimization strategies.

[0044] The pre-trained scene feature-driving parameter mapping model in AI collaborative control module 3 was trained on a training dataset of over 100,000 samples. The training dataset covers variables including: brightness range of 0 to 255 cd / m², object motion speed range of 0 to 5 m / s, shooting distance range of 0.1 m to 10 m, and different environmental temperature and vibration conditions. This model ensures the algorithm has broad adaptability and high accuracy across multiple scenes and conditions.

[0045] The dynamic compensation algorithm is configured to be based on the temperature change monitored by the miniature fiber Bragg grating sensor. Calculate a compensation parameter And based on compensation parameters Perform at least one of the following compensation operations: Compensation is performed on the current of the Helmholtz coil or on the spacing of the shot groups; among which, the compensation parameters... Defined by the following formula: ; Where A is the compensation amplitude scaling factor, B is the error function scaling factor, C is the Bessel function scaling factor, and D is... Function parameter scaling factor, E is logarithmic The weighting coefficients of the product, F is the scaling factor for the logarithmic function parameters, and G is... The scaling factor for the function parameters, H, is Function parameter offset, The change in temperature Let be the error function. It is a first-order Bessel function. for function, It is the natural logarithm function. for Function; and, the current compensation ratio for the Helmholtz coil. Compensation amount for the spacing of the storyboard group ,in and As a preset constant, The value ranges from 0.05 to 0.10. The value ranges from 0.02 to 0.05.

[0046] Personalized storage module 5 uses NOR flash memory to store user-defined parameters, historical usage data, and optimized mapping models. The AI ​​decision-making algorithm supports incremental learning, enabling iterative optimization of the mapping model based on historical usage data, with each optimization improving model accuracy by 0.5% to 1%.

[0047] The collaborative control closed loop is configured to operate according to the following process: During the sensing phase, data from image sensors, miniature Hall sensors, miniature fiber Bragg grating sensors, temperature and humidity sensors, and triaxial accelerometers are collected simultaneously. During the decision-making phase, the collected data is processed through AI decision-making algorithms, and driving parameters and compensation parameters are output. During the execution phase, the main MCU controls the Helmholtz coil and eddy current damper according to the driving parameters and compensation parameters, driving the split-camera group to perform multi-axis fine-tuning. During the feedback phase, the status data and imaging effect data of the lens group after adjustment are obtained through miniature Hall sensors and image sensors; During the optimization phase, the AI ​​decision algorithm compares the target imaging effect with the actual imaging effect. If the deviation exceeds 5%, the driving parameters are optimized a second time and the process is re-executed until the deviation does not exceed 5%. The total response time of the entire process from the first execution to finally meeting the deviation requirement is controlled within 50ms, and the number of iterations of the second optimization does not exceed 3.

[0048] Working principle of Example 3: Sensing stage: The image signal processor extracts image features, the miniature Hall sensor 23 and the fiber optic grating sensor collect motion state and temperature change at a sampling frequency of 1kHz, and the environmental sensing module 4 collects temperature, humidity and vibration information.

[0049] Decision-making phase: The AI ​​decision-making algorithm calls the pre-trained mapping model, integrates the above multi-source data, dynamically outputs driving parameters according to scene conditions, and combines dynamic compensation algorithms to... Perform compensation calculations and generate and .

[0050] Execution phase: Based on the driving parameters and compensation parameters, the main MCU precisely controls the current of the Helmholtz coil 21 and the damping coefficient of the eddy current damper 22 to ensure that the split-screen group 1 moves stably within the suspension gap range, and eliminates the influence of thermal effects through compensation.

[0051] Feedback phase: The Hall sensor and image signal processor collect the adjusted position information and imaging effect of the split-lens group 1 and transmit it to the AI ​​collaborative control module 3.

[0052] Optimization Phase: The AI ​​decision algorithm compares the target imaging effect with the actual imaging result. If the deviation exceeds 5%, it completes up to three iterations of optimization within 50ms until the deviation is controlled within the threshold range. Simultaneously, NOR flash memory records the optimized mapping model and data, providing a basis for subsequent incremental learning.

[0053] Technical effects of Example 3: The mapping model trained on a large number of samples ensures that the lens module has broad adaptability and high-precision control capabilities under different lighting, movement speed, shooting distance and environmental conditions.

[0054] The dynamic compensation algorithm, combined with a complex mathematical function model, achieves a highly sensitive response to temperature changes, effectively avoiding imaging deviations caused by thermal drift.

[0055] based on and The current compensation and spacing compensation enable the shot assembly to maintain stable suspension and precise focus even under temperature fluctuations.

[0056] NOR flash memory, combined with AI incremental learning mechanisms, enables the model accuracy to continuously improve with use, meeting personalized and self-evolving needs.

[0057] The response time is controlled within 50ms, and the number of iterations does not exceed 3, ensuring that the system can still quickly obtain stable imaging results in high-speed motion or severe vibration scenarios.

[0058] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A variable-focus optical lens module, characterized in that, include: Module casing; N split-lens groups (1) are coaxially arranged inside the module housing, where N is an integer greater than or equal to 2; each split-lens group (1) includes a lens holder, M optical lenses (11) fixed inside the lens holder, a permanent magnet ring (12) sleeved on the outside of the lens holder, and a miniature fiber optic grating sensor (13) embedded inside the lens holder, where M is an integer greater than or equal to 3; The magnetic levitation multi-degree-of-freedom drive mechanism (2), the number of which corresponds to the split-screen group (1), includes three pairs of Helmholtz coils (21) fixed to the inner wall of the module housing, the eddy current damper (22) disposed inside the Helmholtz coil (21), and the miniature Hall sensor (23) disposed between the Helmholtz coil (21) and the split-screen group (1); The AI ​​collaborative control module (3) includes a hardware layer (31) and an algorithm layer (32). The hardware layer (31) includes a main MCU, an AI acceleration chip and an image signal processor (ISP). The algorithm layer (32) includes a scene perception algorithm, an AI decision-making algorithm and a dynamic compensation algorithm. The environmental sensing module (4) includes a triaxial accelerometer and a temperature and humidity sensor; Personalized storage module (5) is used to store user-defined parameters and scene feature-driving parameter mapping model; The AI ​​collaborative control module (3) is electrically connected to the magnetic levitation multi-degree-of-freedom drive mechanism (2), the environmental perception module (4) and the personalized storage module (5) respectively, and forms a collaborative control closed loop, which is used to control the Helmholtz coil (21) and the eddy current damper (22) based on sensor data to drive the split-camera group (1) to perform multi-axis fine adjustment.

2. The variable zoom optical lens module according to claim 1, characterized in that: The magnetic levitation multi-degree-of-freedom drive mechanism (2) is configured such that the suspension gap of the split-camera group (1) during the drive process is maintained between 0.1 mm and 0.3 mm, and the control accuracy of the suspension gap is ±0.01 mm. The damping coefficient of the eddy current damper (22) can be dynamically adjusted by the AI ​​collaborative control module (3) in the range of 0.01 N·s / m to 0.05 N·s / m.

3. The variable-focus optical lens module according to claim 1, characterized in that: The sampling frequency of both the miniature Hall sensor (23) and the miniature fiber optic grating sensor (13) is not less than 1 kHz; The sampling frequency of the triaxial accelerometer is not less than 100Hz.

4. The variable zoom optical lens module according to claim 1, characterized in that: The pre-trained scene feature-driving parameter mapping model is trained on a training dataset containing more than 100,000 samples. The variables covered by the training dataset include: The brightness range is 0 to 255 cd / m², the object movement speed range is 0 to 5 m / s, the shooting distance range is 0.1 m to 10 m, and different ambient temperature and vibration conditions are also considered.

5. The variable-focus optical lens module according to claim 1, characterized in that: The dynamic compensation algorithm is configured to be based on the temperature change monitored by the miniature fiber Bragg grating sensor. Calculate a compensation parameter And based on the compensation parameters Perform at least one of the following compensation operations: The current of the Helmholtz coil is compensated, or the spacing of the shot group is compensated; wherein, the compensation parameter... Defined by the following formula: ; Where A is the compensation amplitude scaling factor, B is the error function scaling factor, C is the Bessel function scaling factor, and D is... Function parameter scaling factor, E is logarithmic The weighting coefficients of the product, F is the scaling factor for the logarithmic function parameters, and G is... The scaling factor for the function parameters, H, is Function parameter offset, The change in temperature Let be the error function. It is a first-order Bessel function. for function, It is the natural logarithm function. for The function; and the current compensation ratio of the Helmholtz coil is configured as follows: The spacing compensation amount for the shot group is configured as follows: ,in and As a preset constant, The value ranges from 0.05 to 0.

10. The value ranges from 0.02 to 0.

05.

6. The variable zoom optical lens module according to claim 1, characterized in that: The personalized storage module (5) is a NOR flash memory; The AI ​​decision-making algorithm supports incremental learning, which is used to iteratively optimize the scene feature-driving parameter mapping model based on the historical usage data, and improve the model accuracy by 0.5% to 1% after each optimization.

7. The variable-focus optical lens module according to claim 1, characterized in that: The collaborative control closed loop is configured to operate according to the following process: During the sensing phase, data from the image sensor, the miniature Hall sensor (23), the miniature fiber optic grating sensor (13), the temperature and humidity sensor, and the triaxial accelerometer are collected simultaneously. During the decision-making phase, the collected data is fused and processed through the AI ​​decision-making algorithm, and driving parameters and compensation parameters are output. During the execution phase, the main MCU controls the Helmholtz coil (21) and the eddy current damper (22) according to the driving parameters and compensation parameters, and drives the split-camera group (1) to perform multi-axis fine-tuning. During the feedback phase, the adjusted state data and imaging effect data of the split-lens group (1) are obtained through the micro Hall sensor (23) and the image sensor; During the optimization phase, the AI ​​decision algorithm compares the target imaging effect with the actual imaging effect. If the deviation exceeds 5%, the driving parameters are optimized a second time and the algorithm is re-executed until the deviation does not exceed 5%. The total response time of the entire process from the first execution to finally meeting the deviation requirement is controlled within 50ms, and the number of iterations of the second optimization does not exceed 3.

8. The variable-focus optical lens module according to claim 1, characterized in that: The inner wall of the module housing is provided with an electromagnetic shielding layer, flexible light-blocking floss and micro heat dissipation channels, and the outer side is attached with a graphene heat dissipation film. The module shell is made of carbon fiber composite material or high-strength aluminum alloy, and its wall thickness is 0.5mm to 1.2mm. The electromagnetic shielding layer is a copper foil layer plated on the inner wall of the module housing, with a thickness of 0.02 mm to 0.05 mm.

9. An imaging device, comprising a housing, an image sensor, and a processing unit, characterized in that, The imaging device integrates a variable-focus optical lens module as described in any one of claims 1 to 8.

Images