Electronic atomizer chip visual recognition precision calibration system

Abstract

The present invention relates to the field of electronic manufacturing technology, and specifically to an electronic atomizer chip visual recognition precision calibration system, including a multispectral imaging module, an environmental adaptation module, a deep fusion analysis module, a dynamic calibration control module, a feedback optimization module, and an interactive user interface module; wherein: the multispectral imaging module: is used to collect multispectral image data of the chip; the environmental adaptation module: is used to dynamically adjust the light source intensity and imaging parameters of the multispectral imaging module; the deep fusion analysis module: identifies the type, position, and surface micro-defects of the chip; the dynamic calibration control module: is used to adjust the position and angle of the mechanical arm; the feedback optimization module: optimizes the calibration parameters. The present invention significantly improves the accuracy and efficiency of chip calibration by utilizing advanced capsule network algorithms and dynamic control technology, thereby meeting the urgent needs of the modern electronic manufacturing industry for efficient and precise calibration technology.

Description

Technical Field

[0001] The present invention relates to the field of electronic manufacturing technology, and in particular to a visual recognition and precision calibration system for an electronic atomizer chip. Background Art

[0002] In the electronics manufacturing industry, especially in the production process of electronic atomizers, accurate calibration of chips is crucial to ensure the performance of the equipment. Traditional chip calibration technology relies on artificial vision systems or simple automated equipment. These methods are often unable to handle the precise positioning of chips and the detection of tiny defects, especially in application scenarios with high precision requirements, such as the control chip of electronic atomizers, which has extremely high requirements for chip positioning accuracy and angle adjustment. In addition, the automation systems in the existing technology are often unable to adapt to slight changes in the production environment, such as fluctuations in temperature, humidity and light intensity, which further limits the efficiency and reliability of the calibration system.

[0003] There are several major problems with the existing chip calibration system: first, there is a lack of an efficient dynamic adjustment mechanism, which makes it impossible to provide real-time feedback and optimize the calibration process, resulting in insufficient calibration accuracy; second, the automation system has poor adaptability and cannot dynamically adjust the calibration parameters according to different production conditions, affecting production flexibility and chip quality. In addition, the existing system usually does not integrate an effective user interaction interface, making it difficult for operators to effectively monitor and promptly adjust the calibration process. In response to the above problems, there is an urgent need for a high-precision, highly adaptable and user-friendly chip visual recognition precision calibration system. Summary of the invention

[0004] Based on the above objectives, the present invention provides an electronic atomizer chip visual recognition precision calibration system.

[0005] The electronic atomizer chip visual recognition precision calibration system includes a multi-spectral imaging module, an environmental adaptation module, a deep fusion analysis module, a dynamic calibration control module, a feedback optimization module and an interactive user interface module; among which:

[0006] Multispectral imaging module: equipped with light sources of multiple wavelengths and multispectral cameras to collect multispectral image data of the chip;

[0007] Environmental Adaptive Module: connected to the multispectral imaging module, used to monitor real-time environmental data, including light intensity, temperature and humidity, and dynamically adjust the light intensity and imaging parameters of the multispectral imaging module according to the real-time environmental data;

[0008] Deep fusion analysis module: receives and processes image data from the multispectral imaging module, and uses capsule network algorithm for deep learning analysis to identify the type, location and surface micro-defects of the chip;

[0009] Dynamic calibration control module: connected to the deep fusion analysis module, based on the chip type, position and defect information identified by the deep fusion analysis module, the high-precision servo motor is controlled to adjust the position and angle of the robotic arm to achieve accurate placement and calibration of the chip;

[0010] Feedback optimization module: connected to the dynamic calibration control module, collects quality feedback information after calibration, and analyzes it in combination with historical calibration data to automatically optimize calibration parameters; and feeds back the optimization results to the dynamic calibration control module and the environmental adaptation module to continuously improve calibration accuracy and efficiency;

[0011] Interactive User Interface Module: Used to provide an operation interface and real-time calibration feedback display, allowing users to enter calibration parameters, monitor system status, and adjust system settings as needed.

[0012] Optionally, the multispectral imaging module includes a light source unit, a multispectral camera unit, an image acquisition control unit and a data processing unit; wherein:

[0013] Light source unit: equipped with light sources of multiple wavelengths, including ultraviolet light source, visible light source and infrared light source, used to emit light beams of corresponding wavelengths according to control signals to irradiate the chip with multiple angles and multiple spectra;

[0014] Multispectral camera unit: It is equipped with multiple high-sensitivity sensors, including ultraviolet sensors, visible light sensors, and infrared sensors, which are respectively used to capture the reflected light of the chip within their wavelength ranges and generate multispectral image data of the corresponding wavelengths;

[0015] Image acquisition control unit: including a synchronization control circuit and a timing controller, wherein the synchronization control circuit is used to synchronously control the illumination sequence of the light source unit and the image capture sequence of the multi-spectral camera unit, and the timing controller is used to sequentially activate the ultraviolet light source, the visible light source and the infrared light source according to a preset timing signal, and simultaneously trigger the sensors within the corresponding wavelength range to capture images;

[0016] Data processing unit: used to receive the image data generated by the multispectral camera unit and pre-process the data, including denoising and enhancement processing.

[0017] Optionally, the environment adaptation module includes an environment monitoring unit, a data acquisition unit, a control algorithm unit, a light source adjustment unit and an imaging parameter adjustment unit; wherein:

[0018] Environmental monitoring unit: equipped with light intensity sensor, temperature sensor and humidity sensor, used to monitor the light intensity, temperature and humidity data of the environment in real time;

[0019] Data acquisition unit: connected to the environmental monitoring unit to collect and record real-time data from light intensity, temperature and humidity sensors;

[0020] Control algorithm unit: equipped with an adaptive control algorithm for dynamically calculating and adjusting the light source intensity and imaging parameters of the multispectral imaging module according to the collected environmental data;

[0021] Light source adjustment unit: connected to the control algorithm unit, used to receive the adjustment instruction calculated by the control algorithm unit and dynamically adjust the light source intensity of the multi-spectral imaging module;

[0022] Imaging parameter adjustment unit: connected to the control algorithm unit, used to receive instructions from the control algorithm unit and adjust the imaging parameters of the multispectral imaging module, including exposure time, aperture size and sensor gain.

[0023] Optionally, the control algorithm unit includes:

[0024] Data normalization: Normalize the light intensity L, temperature T and humidity H data collected from the environmental monitoring unit to eliminate dimension and range differences. The formula is:

[0025] Among them, L min and L max are the minimum and maximum preset values ​​of light intensity, T min and T max are the minimum and maximum preset values ​​of temperature, H min and H max are the minimum and maximum preset values ​​of humidity, L′ is the normalized light intensity, T′ is the normalized temperature, and H′ is the normalized humidity;

[0026] Parameter weight allocation: According to experimental data and historical adjustment effects, weights w are allocated to light intensity, temperature and humidity L ,w T ,w H , used to define the influence of each parameter in the adjustment process;

[0027] Light source intensity adjustment calculation: Calculate the value of the adjusted light source intensity I. The specific formula is:

[0028] I=I0×(1+w L ×L′+w T ×f(T′)+w H ×g(H′)), where I0 is the reference light source intensity, f(T′) and g(H′) are functions of the effects of temperature and humidity on the light source intensity adjustment;

[0029] Imaging parameter adjustment calculation: Calculate the adjusted exposure time E and aperture size A, the formula is:

[0030] E=E0×(1-w L × L′); A=A0×(1+w H ×g(H′)); where E0 is the reference exposure time and A0 is the reference aperture size.

[0031] Optionally, the deep fusion analysis module includes a data receiving unit, a preprocessing unit, a capsule network unit, a feature extraction unit and an identification and detection unit; wherein:

[0032] Data receiving unit: used for receiving multispectral image data from the multispectral imaging module;

[0033] Preprocessing unit: used to denoise and normalize the received multispectral image data to prepare for subsequent deep learning analysis;

[0034] Capsule network unit: It is equipped with a capsule network structure and is used to perform deep learning analysis on preprocessed image data to capture features and hierarchical information in the image.

[0035] Feature extraction unit: Based on the capsule network unit, the spatial hierarchical features in the image are extracted through the capsule layer to retain the chip's posture information;

[0036] Identification and detection unit: Utilize the dynamic routing mechanism in the capsule network and the extracted feature vector to identify the type of chip, calculate the position of the chip in the image, detect tiny defects on the surface of the chip, and generate corresponding identification and defect information.

[0037] Optionally, the capsule network unit includes:

[0038] Initial convolution layer: The preprocessed image data is input into the initial convolution layer, and low-level features are extracted through convolution operation. The formula is: F p,q,r =Σ i,j I p+i,q+j ·K i,j,r , where I p+i,q+j represents the pixel value of the input image at position (p+i, q+j), K i,j,r represents the weight of the rth convolution kernel at position (i, j), F p,q,r Represents the rth eigenvalue of the convolution output feature map at position (p, q), where p and q are pixel position indexes, and i and j are position indexes within the convolution kernel;

[0039] Primary capsule layer: The feature map output by the convolutional layer is input into the primary capsule layer. Each primary capsule calculates the feature vector through a nonlinear activation function. The formula is: Among them, s k is the sum vector input to the capsule, vk is the output feature vector, k represents the capsule index;

[0040] Dynamic routing mechanism: The feature vector output by the primary capsule layer is passed to the advanced capsule layer through a dynamic routing mechanism, and the weights are dynamically adjusted to optimize the feature representation. The formula is: ij ←b ij +v j ·u j|i , where v j is the output vector of capsule j after being processed by the activation function, u j|i is the vector that capsule i predicts and sends to capsule j, b ij is the routing coefficient between capsule i and capsule j, which is continuously updated through an iterative process to reflect the connection strength between capsule i and capsule j.

[0041] Optionally, the identification and detection unit includes:

[0042] Feature vector classification: classify the feature vectors obtained from the feature extraction unit to identify the type of chip. Specifically, the dynamic routing mechanism in the capsule network is used to classify the feature vector v output by the capsule layer. k , calculate the activation value a of the capsule k , the formula is: k =softmax(||v k ||), where v k represents the feature vector output by the kth capsule, a k Indicates the probability of existence of the feature, which is the activation value;

[0043] Position calculation: The output feature vector of the capsule network is used to calculate the position of the chip in the image. By analyzing the direction and length of the feature vector, the specific position coordinates x and y of the chip are obtained. The formula is:

[0044] (x, y) = ∑ k (v k ·w k ), where w k is the same as the eigenvector v k The corresponding position weight, x and y represent the position coordinates of the chip in the image;

[0045] Defect detection: By analyzing the feature vectors in the capsule network, tiny defects on the chip surface are detected, and defect information is generated based on the feature vectors of the defects. The formula is: D = Σ k (d k ·m k ), where d k is the defect feature vector corresponding to the kth capsule, m k is the defect feature vector d kThe corresponding weight,D,represents the generated defect information;

[0046] Information generation: The identified chip type, location and defect information are integrated to generate the final identification and defect report.

[0047] Optionally, the dynamic calibration control module includes a data receiving unit, a parameter calculation unit, a servo control unit and a real-time feedback unit; wherein:

[0048] Data receiving unit: used to receive chip type T, position x, y and defect information D from the deep fusion analysis module;

[0049] Parameter calculation unit: Based on the received information, the movement path and angle adjustment of the robot arm are calculated. Using the received chip position and defect information, the rotation angle and movement distance of each joint of the robot arm are calculated to locate the correct placement of the chip. The formula is:

[0050] Among them, x and y are the target position coordinates of the chip, x i and i is the coordinate of the current position of the robot, θ i is the rotation angle of the i-th joint, d i is the distance the robot moves, arctan is the inverse tangent function;

[0051] Servo control unit: According to the parameters calculated by the parameter calculation unit, the high-precision servo motor is controlled to adjust the position and angle of the robot arm. The control signal of the servo motor is determined by the following formula:

[0052]

[0053] in, It is the control signal for the servo motor to control the joint angle of the robot arm. K is the control signal of the servo motor to control the extension distance of the robot arm. p is the proportional gain coefficient, θ current and d current is the current angle and distance;

[0054] Real-time feedback unit: used to monitor the placement status of the chip and the movement status of the robot arm in real time during the adjustment process. If position or angle deviation is found, the control parameters are adjusted immediately for correction. The specific adjustment is made through the following formula: Δθ=θ target -θ measured ; Δd=d target -d measured , where θ tar get and d tar get is the target angle and distance, θ measured and dmeasured are the actual measured angle and distance, Δθ and Δd are the deviations.

[0055] Optionally, the feedback optimization module includes a quality feedback collection unit, a historical data storage unit, a data analysis unit, and an optimization result feedback unit; wherein:

[0056] Feedback data collection unit: used to collect quality feedback information after calibration, including the final position, angle and calibration accuracy of the chip;

[0057] Historical data storage unit: used to store previous calibration data, including parameters, calibration results and related environmental data of each calibration;

[0058] Data analysis unit: used to combine quality feedback information and historical calibration data to analyze and identify factors affecting calibration accuracy and generate optimization suggestions. The specific steps include:

[0059] Step 1: Normalize the current quality feedback information and historical calibration data to eliminate dimensional differences;

[0060] Step 2: Analyze the influence of different calibration parameters on the calibration results through linear regression or nonlinear regression model. The formula is: Q = β0 + β1P1 + β2P2 + ... + β n P n +∈, where Q is the calibration quality, P is the i-th calibration parameter, β i is the regression coefficient, ∈ is the error term, β0 is the reference value, β1,β2,…,β n are the regression coefficients of the calibration parameters, P1, P2, …, P n are the various calibration parameters;

[0061] Step 3: Based on the regression analysis results, the calibration parameters are adjusted using the gradient descent optimization algorithm. The formula is: in, and are the new and old calibration parameter values, respectively, and α is the learning rate;

[0062] Optimization result feedback unit: Feedback the optimization results to the dynamic calibration control module and the environmental adaptation module, including updating the calibration parameters to adjust the position and angle of the robotic arm, and adjusting the light source intensity and imaging parameters of the multispectral imaging module.

[0063] Optionally, the interactive user interface module includes an operation interface unit, a real-time feedback display unit, a parameter input unit, a system monitoring unit and a setting adjustment unit; wherein:

[0064] Operation interface unit: used to provide a user operation interface, allowing the user to input calibration parameters, start or stop the calibration process, and select the calibration mode;

[0065] Real-time feedback display unit: used to display real-time data and feedback information during the calibration process, including the current chip position, angle, calibration accuracy, and detected defect information;

[0066] Parameter input unit: supports users to input calibration parameters, including target position x, y, rotation angle θ;

[0067] System monitoring unit: used to monitor the operating status of the system, including the motion status of the robot arm, the working status of the servo motor, and the parameter adjustment status of the environmental adaptive module;

[0068] Setting adjustment unit: used to adjust system settings according to user operations and feedback information, including fine-tuning of calibration parameters, response speed adjustment of servo motors, and parameter adjustment of environmental adaptation modules.

[0069] Beneficial effects of the present invention:

[0070] The present invention significantly improves the accuracy and efficiency of chip calibration by integrating advanced capsule network algorithms and dynamic calibration control technology. The capsule network algorithm can effectively capture the spatial hierarchical features and posture information of the chip, which is crucial for accurately locating the position and angle of the chip. In addition, the dynamic calibration control module can automatically adjust the movement and rotation of the robot arm according to real-time feedback information to ensure the precise placement of the chip on the production line, thereby reducing the error rate and time delay in production and improving production efficiency.

[0071] The present invention further enhances the adaptability of the calibration process through the environmental adaptation module, and can adjust the calibration parameters according to real-time environmental changes, thereby ensuring that high-quality calibration results can be maintained in various production environments. At the same time, the interactive user interface module provides an intuitive and easy-to-operate user interface, allowing operators to easily input calibration parameters, monitor the calibration process and adjust system settings, greatly improving the user's operating experience and the controllability of the system. These features enable the present invention to not only adapt to complex and changeable production needs, but also provide more humane and intelligent user interaction, thereby meeting the urgent needs of the modern electronic manufacturing industry for efficient and accurate calibration technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0073] Figure 1 A schematic diagram of a visual recognition precision calibration system according to an embodiment of the present invention;

[0074] Figure 2 Schematic diagram of a deep fusion analysis module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0075] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0076] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0077] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0078] like Figure 1-Figure 2 As shown, the electronic atomizer chip visual recognition precision calibration system includes a multi-spectral imaging module, an environmental adaptation module, a deep fusion analysis module, a dynamic calibration control module, a feedback optimization module and an interactive user interface module; wherein:

[0079] Multispectral imaging module: equipped with light sources of multiple wavelengths and multispectral cameras, used to collect multispectral image data of the chip and transmit it to the deep fusion analysis module;

[0080] Environmental Adaptive Module: Connected with the multispectral imaging module, it is used to monitor real-time environmental data, including light intensity, temperature and humidity, and dynamically adjust the light intensity and imaging parameters of the multispectral imaging module according to the real-time environmental data to ensure the consistency and accuracy of the multispectral image data;

[0081] Deep fusion analysis module: receives and processes image data from the multispectral imaging module, and uses the capsule network algorithm to perform deep learning analysis to identify the type, location and surface defects of the chip. The analysis results are used to guide the adjustment of the dynamic calibration control module;

[0082] Dynamic calibration control module: connected to the deep fusion analysis module, based on the chip type, position and defect information identified by the deep fusion analysis module, the high-precision servo motor is controlled to adjust the position and angle of the robotic arm to achieve accurate placement and calibration of the chip;

[0083] Feedback optimization module: connected to the dynamic calibration control module, collects quality feedback information after calibration, and analyzes it in combination with historical calibration data to automatically optimize calibration parameters; and feeds back the optimization results to the dynamic calibration control module and the environmental adaptation module to continuously improve calibration accuracy and efficiency;

[0084] Interactive User Interface Module: Used to provide an operation interface and real-time calibration feedback display, allowing users to enter calibration parameters, monitor system status, and adjust system settings as needed.

[0085] The multispectral imaging module includes a light source unit, a multispectral camera unit, an image acquisition control unit and a data processing unit; wherein:

[0086] Light source unit: equipped with light sources of multiple wavelengths, including ultraviolet light source (wavelength range of 200-400nm), visible light source (wavelength range of 400-700nm) and infrared light source (wavelength range of 700-1000nm), used to emit light beams of corresponding wavelengths according to control signals to irradiate the chip with multiple angles and multiple spectra;

[0087] Multispectral camera unit: It is equipped with multiple high-sensitivity sensors, including ultraviolet sensors, visible light sensors, and infrared sensors, which are respectively used to capture the reflected light of the chip within their wavelength ranges and generate multispectral image data of the corresponding wavelengths;

[0088] Image acquisition control unit: including a synchronization control circuit and a timing controller. The synchronization control circuit is used to synchronously control the illumination sequence of the light source unit and the image capture sequence of the multi-spectral camera unit. The timing controller is used to activate the ultraviolet light source, the visible light source and the infrared light source in sequence according to a preset timing signal, and simultaneously trigger the sensors within the corresponding wavelength range to capture images, thereby ensuring the timing consistency of the image data of each wavelength;

[0089] Data processing unit: used to receive the image data generated by the multispectral camera unit and pre-process the data, including denoising and enhancement processing, to improve the quality and accuracy of the image data; through the design of the above units, the multispectral imaging module can fully illuminate the chip and capture high-precision images at different wavelengths, ensuring that the acquired multispectral image data has high quality and high consistency. This design improves the ability to identify tiny defects and material components on the chip surface, enhances the robustness and adaptability of the system in complex environments, and significantly improves calibration accuracy and production efficiency.

[0090] The environment adaptation module includes an environment monitoring unit, a data acquisition unit, a control algorithm unit, a light source adjustment unit and an imaging parameter adjustment unit; wherein:

[0091] Environmental monitoring unit: equipped with light intensity sensor, temperature sensor and humidity sensor, used to monitor the light intensity, temperature and humidity data of the environment in real time;

[0092] Data acquisition unit: connected to the environmental monitoring unit to collect and record real-time data from light intensity, temperature and humidity sensors;

[0093] Control algorithm unit: equipped with an adaptive control algorithm for dynamically calculating and adjusting the light source intensity and imaging parameters of the multispectral imaging module according to the collected environmental data;

[0094] Light source adjustment unit: connected to the control algorithm unit, used to receive the adjustment instruction calculated by the control algorithm unit and dynamically adjust the light source intensity of the multi-spectral imaging module;

[0095] Imaging parameter adjustment unit: connected to the control algorithm unit, used to receive instructions from the control algorithm unit and adjust the imaging parameters of the multispectral imaging module, including exposure time, aperture size and sensor gain;

[0096] Through the design of the above units, the environmental adaptation module can monitor the ambient light intensity, temperature and humidity in real time, and dynamically adjust the light source intensity and imaging parameters of the multispectral imaging module according to these data, ensuring that high-quality and highly consistent multispectral image data can be obtained under different environmental conditions. This design significantly enhances the system's environmental adaptability and imaging stability, improves the accuracy of identifying tiny defects and material composition on the chip surface, and improves the reliability and efficiency of the overall calibration system.

[0097] The control algorithm unit includes:

[0098] Data normalization: Normalize the light intensity L, temperature T and humidity H data collected from the environmental monitoring unit to eliminate dimension and range differences. The formula is:

[0099] Among them, L min and L max are the minimum and maximum preset values ​​of light intensity, T min and T max are the minimum and maximum preset values ​​of temperature, H min and H max are the minimum and maximum preset values ​​of humidity, L′ is the normalized light intensity, T′ is the normalized temperature, and H′ is the normalized humidity;

[0100] Parameter weight allocation: According to experimental data and historical adjustment effects, weights w are allocated to light intensity, temperature and humidity L ,w T ,w H , used to define the influence of each parameter in the adjustment process;

[0101] Light source intensity adjustment calculation: Calculate the value of the adjusted light source intensity I. The specific formula is:

[0102] I=I0×(1+w L ×L′+w T ×f(T′)+w H ×g(H′)), where I0 is the reference light source intensity, f(T′) and g(H′) are functions of the effects of temperature and humidity on the light source intensity adjustment, which can be linear or nonlinear functions and are customized according to actual application requirements;

[0103] Imaging parameter adjustment calculation: Calculate the adjusted exposure time E and aperture size A, the formula is:

[0104] E=E0×(1-w L × L′); A=A0×(1+w H ×g(H′)); wherein, E0 is the reference exposure time, A0 is the reference aperture size: finally, according to the light source intensity adjustment calculation and the imaging parameter adjustment calculation, the light source adjustment unit and the imaging parameter adjustment unit perform corresponding adjustments; the above-mentioned adaptive control algorithm can effectively adapt to various environmental changes and improve the quality and consistency of the image by accurately calculating and dynamically adjusting the light source intensity and imaging parameters, which not only improves the accuracy of image analysis, but also enhances the flexibility and reliability of the entire system, and significantly improves the efficiency and accuracy of chip surface defect detection.

[0105] The deep fusion analysis module includes a data receiving unit, a preprocessing unit, a capsule network unit, a feature extraction unit, and a recognition and detection unit; wherein:

[0106] Data receiving unit: used to receive multispectral image data from the multispectral imaging module and perform preliminary data integration to ensure data integrity and consistency;

[0107] Preprocessing unit: used to perform denoising and normalization on the received multispectral image data to improve the quality of the image data and prepare for subsequent deep learning analysis;

[0108] Capsule network unit: It is equipped with a capsule network structure and is used to perform deep learning analysis on preprocessed image data to capture features and hierarchical information in the image.

[0109] Feature extraction unit: Based on the capsule network unit, the spatial hierarchical features in the image are extracted through the capsule layer to retain the chip's posture information;

[0110] Identification and detection unit: Utilize the dynamic routing mechanism in the capsule network, combined with the extracted feature vectors, to identify the type of chip, calculate the position of the chip in the image, detect tiny defects on the chip surface, and generate corresponding identification and defect information; through the above design, the deep fusion analysis module can efficiently process and analyze multispectral image data, use the capsule network algorithm for deep learning analysis, effectively identify the type, position and tiny surface defects of the chip, and the capsule network structure can retain the spatial hierarchical information in the image and improve the accuracy of identification and detection. This design significantly enhances the detection capability and accuracy of the system and improves the overall performance of chip visual recognition and calibration.

[0111] The capsule network unit includes:

[0112] Initial convolution layer: The preprocessed image data is input into the initial convolution layer, and low-level features are extracted through convolution operation. The formula is: F p,q,r =∑ i,j I p+i,q+j ·K i,j,r , where I p+i,q+j represents the pixel value of the input image at position (p+i, q+j), K i,j,r represents the weight of the rth convolution kernel at position (i, j), F p,q,r Represents the rth eigenvalue of the convolution output feature map at position (p, q), where p and q are pixel position indexes, and i and j are position indexes within the convolution kernel;

[0113] Primary capsule layer: The feature map output by the convolutional layer is input into the primary capsule layer. Each primary capsule calculates the feature vector through a nonlinear activation function. The formula is: Among them, s k is the sum vector input to the capsule, v k is the output feature vector, k represents the capsule index;

[0114] Dynamic routing mechanism: The feature vector output by the primary capsule layer is passed to the advanced capsule layer through a dynamic routing mechanism, and the weights are dynamically adjusted to optimize the feature representation. The formula is: ij ←b ij +v j ·u j|i , where v j is the output vector of capsule j after being processed by the activation function, u j|i is the vector that capsule i predicts and sends to capsule j, b ij is the routing coefficient between capsule i and capsule j, which is continuously updated through an iterative process to reflect the connection strength between capsule i and capsule j. Through the above design, the capsule network unit can effectively extract and analyze the complex features of multispectral image data, capture the spatial hierarchy and posture information in the image, and the dynamic routing mechanism optimizes the transmission and representation of feature vectors, enhancing the accuracy of recognition and detection. This design significantly improves the deep learning analysis capability and overall performance of the system, ensuring the high accuracy and reliability of chip visual recognition and calibration.

[0115] The identification and detection unit includes:

[0116] Feature vector classification: classify the feature vectors obtained from the feature extraction unit to identify the type of chip. Specifically, the dynamic routing mechanism in the capsule network is used to classify the feature vector v output by the capsule layer. k , calculate the activation value a of the capsule k , the formula is: k =softmax(||v k ||), where v k represents the feature vector output by the kth capsule, which represents the specific characteristics of the chip or its part, a k Indicates the probability of existence of the feature, which is the activation value;

[0117] Position calculation: The output feature vector of the capsule network is used to calculate the position of the chip in the image. By analyzing the direction and length of the feature vector, the specific position coordinates x and y of the chip are obtained. The formula is:

[0118] (x, y) = ∑ k (v k ·w k ), where w k is the same as the eigenvector v k The corresponding position weight is used to calculate the coefficient of the position, where x and y represent the position coordinates of the chip in the image;

[0119] Defect detection: By analyzing the feature vectors in the capsule network, tiny defects on the chip surface are detected, and defect information is generated based on the feature vectors of the defects. The formula is: D = ∑k (d k ·m k ), where d k is the defect feature vector corresponding to the kth capsule, and mk is the defect feature vector d k The corresponding weight is used to calculate the coefficient of defect information, and D represents the generated defect information;

[0120] Information generation: Integrate the identified chip type, location and defect information to generate the final identification and defect report;

[0121] Through the above design, the recognition and detection unit can efficiently utilize the dynamic routing mechanism and feature vectors in the capsule network to accurately identify the type, location and surface defects of the chip. The dynamic routing mechanism enhances the transmission and representation of feature vectors, making the recognition and detection results more accurate and reliable. This design significantly improves the recognition and detection capabilities of the system, ensuring the high accuracy and reliability of chip visual recognition and calibration.

[0122] The dynamic calibration control module includes a data receiving unit, a parameter calculation unit, a servo control unit and a real-time feedback unit; wherein:

[0123] Data receiving unit: used to receive chip type T, position x, y and defect information D from the deep fusion analysis module, ensuring that these data contain necessary details, such as specific coordinate positions and the type and degree of identified defects;

[0124] Parameter calculation unit: Based on the received information, the movement path and angle adjustment of the robot arm are calculated. Using the received chip position and defect information, the rotation angle and movement distance of each joint of the robot arm are calculated to locate the correct placement of the chip. The formula is:

[0125] Among them, x and y are the target position coordinates of the chip, x i and i is the coordinate of the current position of the robot, θ i is the rotation angle of the i-th joint, d i is the distance the robot moves, and arctan is the inverse tangent function, which is used to calculate the angle to which the joints of the robot need to rotate;

[0126] Servo control unit: According to the parameters calculated by the parameter calculation unit, the high-precision servo motor is controlled to adjust the position and angle of the robot arm. The control signal of the servo motor is determined by the following formula:

[0127]

[0128] in, It is the control signal for the servo motor to control the joint angle of the robot arm. K is the control signal of the servo motor to control the extension distance of the robot arm. p is the proportional gain coefficient, θ current and d current is the current angle and distance;

[0129] Real-time feedback unit: used to monitor the placement status of the chip and the movement status of the robot arm in real time during the adjustment process. If position or angle deviation is found, the control parameters are adjusted immediately for correction. The specific adjustment is made through the following formula: Δθ=θ target -θ measured ; Δd=d target -d measured , where θ tar get and d tar get is the target angle and distance, θ measured and d measured are the actual measured angle and distance, Δθ and Δd are the deviations; through the precise adjustment and real-time feedback mechanism of the dynamic calibration control module, the system can efficiently and accurately adjust the position and angle of the chip according to the identified chip type, position and defect information. This fine control not only improves the accuracy of calibration, but also significantly enhances production efficiency and chip quality, ensuring the reliability and effectiveness of the calibration process.

[0130] The feedback optimization module includes a quality feedback collection unit, a historical data storage unit, a data analysis unit, and an optimization result feedback unit; wherein:

[0131] Feedback data collection unit: used to collect quality feedback information after calibration, including the final position, angle and calibration accuracy of the chip;

[0132] Historical data storage unit: used to store previous calibration data, including parameters, calibration results and related environmental data of each calibration;

[0133] Data analysis unit: used to combine quality feedback information and historical calibration data to analyze and identify factors affecting calibration accuracy and generate optimization suggestions. The specific steps include:

[0134] Step 1: Normalize the current quality feedback information and historical calibration data to eliminate dimensional differences;

[0135] Step 2: Analyze the influence of different calibration parameters on the calibration results through linear regression or nonlinear regression model. The formula is: Q = β0 + β1P1 + β2P2 + ... + β n P n +∈, where Q is the calibration quality; β0 is the constant term (intercept term), which means that in all calibration parameters P1 to P nWhen the value of is zero, the initial value of the calibration quality Q, which is a fixed baseline value used to calibrate the benchmark of system performance; β1 to β n are regression coefficients, which represent each calibration parameter P1 to P n The impact on the calibration quality Q, specifically, β i Indicates that when the calibration parameter P i The change in calibration mass Q when one unit is added, other parameters remain unchanged, these coefficients are learned from historical data through regression analysis; P1 to P n are calibration parameters, representing different calibration variables in the dynamic calibration control system, such as the movement distance, rotation angle, response time of the servo motor, etc. of the robot arm; each P i Corresponding to a specific adjustable parameter in a system, it affects the calibration quality Q; ∈ is the error term, which represents the difference between the actual calibration quality Q and the calibration quality predicted by the regression model. This term captures the random errors or unknown factors that the model cannot explain.

[0136] Step 3: Based on the regression analysis results, the calibration parameters are adjusted using the gradient descent optimization algorithm. The formula is: in, and are the new and old calibration parameter values, respectively, and α is the learning rate;

[0137] Optimization result feedback unit: Feedback the optimization results to the dynamic calibration control module and the environment adaptation module, including updating the calibration parameters to adjust the position and angle of the robot arm, and adjusting the light source intensity and imaging parameters of the multispectral imaging module to adapt to the optimized calibration requirements;

[0138] The feedback data collection unit obtains the quality feedback information after calibration from the visual system and stores this information in the historical data storage unit. The data analysis unit identifies the main factors affecting the calibration results through normalization processing and regression analysis, and uses the optimization algorithm to adjust the calibration parameters. The optimization result feedback unit feeds back the new calibration parameters to the dynamic calibration control module and the environmental adaptation module to ensure that the system continuously improves the calibration accuracy and efficiency.

[0139] The interactive user interface module includes an operation interface unit, a real-time feedback display unit, a parameter input unit, a system monitoring unit, and a setting adjustment unit; wherein:

[0140] Operation interface unit: used to provide a user operation interface, allowing the user to input calibration parameters, start or stop the calibration process, and select the calibration mode;

[0141] Real-time feedback display unit: used to display real-time data and feedback information during the calibration process, including the current chip position, angle, calibration accuracy, and detected defect information;

[0142] Parameter input unit: supports users to input calibration parameters, including target position x, y, rotation angle θ, and other related calibration parameters;

[0143] System monitoring unit: used to monitor the operating status of the system, including the motion status of the robot arm, the working status of the servo motor, and the parameter adjustment status of the environmental adaptive module;

[0144] Setting adjustment unit: used to adjust system settings according to user operations and feedback information, including fine-tuning of calibration parameters, response speed adjustment of servo motors, and parameter adjustment of environmental adaptive modules;

[0145] The operation interface unit provides a user operation interface through a touch screen or other input devices, so that the user can easily input calibration parameters and control the calibration process. The real-time feedback display unit displays the real-time data and feedback information collected during the calibration process, so that the user can instantly understand the current calibration status. The parameter input unit receives the calibration parameters input by the user and passes these parameters to the relevant modules for processing. The system monitoring unit monitors the operating status of the entire system to ensure the normal operation of each unit. The setting adjustment unit dynamically adjusts the system settings according to the user's operation and feedback information to ensure the accuracy and efficiency of the calibration process.

[0146] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0147] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. Electronic atomizer chip visual recognition precision calibration system, characterized by: It includes a multispectral imaging module, an environmental adaptation module, a deep fusion analysis module, a dynamic calibration control module, a feedback optimization module and an interactive user interface module; among which: Multispectral imaging module: equipped with light sources of multiple wavelengths and multispectral cameras to collect multispectral image data of the chip; Environmental Adaptive Module: connected to the multispectral imaging module, used to monitor real-time environmental data, including light intensity, temperature and humidity, and dynamically adjust the light intensity and imaging parameters of the multispectral imaging module according to the real-time environmental data; The environment adaptation module includes an environment monitoring unit, a data acquisition unit, a control algorithm unit, a light source adjustment unit and an imaging parameter adjustment unit; wherein: Environmental monitoring unit: equipped with light intensity sensor, temperature sensor and humidity sensor, used to monitor the light intensity, temperature and humidity data of the environment in real time; Data acquisition unit: connected to the environmental monitoring unit to collect and record real-time data from light intensity, temperature and humidity sensors; Control algorithm unit: equipped with an adaptive control algorithm for dynamically calculating and adjusting the light source intensity and imaging parameters of the multispectral imaging module according to the collected environmental data; Light source adjustment unit: connected to the control algorithm unit, used to receive the adjustment instruction calculated by the control algorithm unit and dynamically adjust the light source intensity of the multi-spectral imaging module; Imaging parameter adjustment unit: connected to the control algorithm unit, used to receive instructions from the control algorithm unit and adjust the imaging parameters of the multispectral imaging module, including exposure time, aperture size and sensor gain; The control algorithm unit comprises: Data normalization: Normalize the light intensity L, temperature T and humidity H data collected from the environmental monitoring unit to eliminate dimension and range differences. The formula is: Among them, L min and L max are the minimum and maximum preset values ​​of light intensity, T min and T max are the minimum and maximum preset values ​​of temperature, H min and H max are the minimum and maximum preset values ​​of humidity, L′ is the normalized light intensity, T′ is the normalized temperature, and H′ is the normalized humidity; Parameter weight allocation: According to experimental data and historical adjustment effects, weights w are allocated to light intensity, temperature and humidity L ,w T ,w H , used to define the influence of each parameter in the adjustment process; Light source intensity adjustment calculation: Calculate the value of the adjusted light source intensity I. The specific formula is: I=I0×(1+w L ×L′+w T ×f(T′)+w H ×g(H′)), where I0 is the reference light source intensity, f(T′) and g(H′) are functions of the effects of temperature and humidity on the light source intensity adjustment; Imaging parameter adjustment calculation: Calculate the adjusted exposure time E and aperture size A, the formula is: E=E0×(1-w L × L′); A=A0×(1+w H ×g(H′)); where E0 is the reference exposure time and A0 is the reference aperture size; Deep fusion analysis module: receives and processes image data from the multispectral imaging module, and uses capsule network algorithm for deep learning analysis to identify the type, location and surface micro-defects of the chip; Dynamic calibration control module: connected to the deep fusion analysis module, based on the chip type, position and defect information identified by the deep fusion analysis module, the high-precision servo motor is controlled to adjust the position and angle of the robotic arm to achieve accurate placement and calibration of the chip; The dynamic calibration control module includes a data receiving unit, a parameter calculation unit, a servo control unit and a real-time feedback unit; wherein: Data receiving unit: used to receive chip type T, position x, y and defect information D from the deep fusion analysis module; Parameter calculation unit: Based on the received information, the movement path and angle adjustment of the robot arm are calculated. Using the received chip position and defect information, the rotation angle and movement distance of each joint of the robot arm are calculated to locate the correct placement of the chip. The formula is: Among them, x and y are the target position coordinates of the chip, x i and i is the coordinate of the current position of the robot, θ i is the rotation angle of the i-th joint, d i is the distance the robot moves, arctan is the inverse tangent function; Servo control unit: According to the parameters calculated by the parameter calculation unit, the high-precision servo motor is controlled to adjust the position and angle of the robot arm. The control signal of the servo motor is determined by the following formula: in, It is the control signal for the servo motor to control the joint angle of the robot arm. K is the control signal of the servo motor to control the extension distance of the robot arm. p is the proportional gain coefficient, θ current and d current is the current angle and distance; Real-time feedback unit: used to monitor the placement status of the chip and the movement status of the robot arm in real time during the adjustment process. If position or angle deviation is found, the control parameters are adjusted immediately for correction. The specific adjustment is made through the following formula: Δθ=θ target -θ measured ; Δd=d target -d measured , where θ target and d target is the target angle and distance, θ measured and d measured are the angle and distance actually measured, Δθ and Δd are the deviations; Feedback optimization module: connected to the dynamic calibration control module, collects quality feedback information after calibration, and analyzes it in combination with historical calibration data to automatically optimize calibration parameters; and feeds back the optimization results to the dynamic calibration control module and the environmental adaptation module to continuously improve calibration accuracy and efficiency; Interactive User Interface Module: Used to provide an operation interface and real-time calibration feedback display, allowing users to enter calibration parameters, monitor system status, and adjust system settings as needed.

2. The electronic atomizer chip visual recognition precision calibration system according to claim 1, characterized in that: The multi-spectral imaging module includes a light source unit, a multi-spectral camera unit, an image acquisition control unit and a data processing unit; wherein: Light source unit: equipped with light sources of multiple wavelengths, including ultraviolet light source, visible light source and infrared light source, used to emit light beams of corresponding wavelengths according to control signals to irradiate the chip with multiple angles and multiple spectra; Multispectral camera unit: It is equipped with multiple high-sensitivity sensors, including ultraviolet sensors, visible light sensors, and infrared sensors, which are respectively used to capture the reflected light of the chip within their wavelength ranges and generate multispectral image data of the corresponding wavelengths; Image acquisition control unit: including a synchronization control circuit and a timing controller, wherein the synchronization control circuit is used to synchronously control the illumination sequence of the light source unit and the image capture sequence of the multi-spectral camera unit, and the timing controller is used to sequentially activate the ultraviolet light source, the visible light source and the infrared light source according to a preset timing signal, and simultaneously trigger the sensors within the corresponding wavelength range to capture images; Data processing unit: used to receive the image data generated by the multispectral camera unit and pre-process the data, including denoising and enhancement processing.

3. The electronic atomizer chip visual recognition precision calibration system according to claim 1, characterized in that: The deep fusion analysis module includes a data receiving unit, a preprocessing unit, a capsule network unit, a feature extraction unit and an identification and detection unit; wherein: Data receiving unit: used for receiving multispectral image data from the multispectral imaging module; Preprocessing unit: used to denoise and normalize the received multispectral image data to prepare for subsequent deep learning analysis; Capsule network unit: It is equipped with a capsule network structure and is used to perform deep learning analysis on preprocessed image data to capture features and hierarchical information in the image. Feature extraction unit: Based on the capsule network unit, the spatial hierarchical features in the image are extracted through the capsule layer to retain the chip's posture information; Identification and detection unit: Utilize the dynamic routing mechanism in the capsule network and the extracted feature vector to identify the type of chip, calculate the position of the chip in the image, detect tiny defects on the surface of the chip, and generate corresponding identification and defect information.

4. The electronic atomizer chip visual recognition precision calibration system according to claim 3, characterized in that: The capsule network unit comprises: Initial convolution layer: The preprocessed image data is input into the initial convolution layer, and low-level features are extracted through convolution operation. The formula is: F p,q,r =∑ i,j I p+i,q+j ·K i,j,r , where I p+i,q+j represents the pixel value of the input image at position (p+i, q+j), K i,j,r represents the weight of the rth convolution kernel at position (i, j), F p,q,r Represents the rth eigenvalue of the convolution output feature map at position (p, q), where p and q are pixel position indexes, and i and j are position indexes within the convolution kernel; Primary capsule layer: The feature map output by the convolutional layer is input into the primary capsule layer. Each primary capsule calculates the feature vector through a nonlinear activation function. The formula is: Among them, s k is the sum vector input to the capsule, v k is the output feature vector, k represents the capsule index; Dynamic routing mechanism: The feature vector output by the primary capsule layer is passed to the advanced capsule layer through a dynamic routing mechanism, and the weights are dynamically adjusted to optimize the feature representation. The formula is: ij ←b ij +v j ·u j|i , where v j is the output vector of capsule j after being processed by the activation function, u j|i is the vector that capsule i predicts and sends to capsule j, b ij is the routing coefficient between capsule i and capsule j, which is continuously updated through an iterative process to reflect the connection strength between capsule i and capsule j.

5. The electronic atomizer chip visual recognition precision calibration system according to claim 4, characterized in that: The identification and detection unit comprises: Feature vector classification: classify the feature vectors obtained from the feature extraction unit to identify the type of chip. Specifically, the dynamic routing mechanism in the capsule network is used to classify the feature vector v output by the capsule layer. k , calculate the activation value a of the capsule k , the formula is: k =softmax(||v k ||), where v k represents the feature vector output by the kth capsule, a k Indicates the probability of existence of the feature, which is the activation value; Position calculation: The output feature vector of the capsule network is used to calculate the position of the chip in the image. By analyzing the direction and length of the feature vector, the specific position coordinates x and y of the chip are obtained. The formula is: (x, y) = ∑ k (v k ·w k ), where w k is the same as the eigenvector v k The corresponding position weight, x and y represent the position coordinates of the chip in the image; Defect detection: By analyzing the feature vectors in the capsule network, tiny defects on the chip surface are detected, and defect information is generated based on the feature vectors of the defects. The formula is: D = ∑ k (d k ·m k ), where d k is the defect feature vector corresponding to the kth capsule, m k is the defect feature vector d k The corresponding weight,D,represents the generated defect information; Information generation: The identified chip type, location and defect information are integrated to generate the final identification and defect report.

6. The electronic atomizer chip visual recognition precision calibration system according to claim 1, characterized in that: The feedback optimization module includes a quality feedback collection unit, a historical data storage unit, a data analysis unit, and an optimization result feedback unit; wherein: Feedback data collection unit: used to collect quality feedback information after calibration, including the final position, angle and calibration accuracy of the chip; Historical data storage unit: used to store previous calibration data, including parameters, calibration results and related environmental data of each calibration; Data analysis unit: used to combine quality feedback information and historical calibration data to analyze and identify factors affecting calibration accuracy and generate optimization suggestions. The specific steps include: Step 1: Normalize the current quality feedback information and historical calibration data to eliminate dimensional differences; Step 2: Analyze the influence of different calibration parameters on the calibration results through linear regression or nonlinear regression model. The formula is: Q = β0 + β1P1 + β2P2 + ... + β n P n +∈, where Q is the calibration quality, P is the i-th calibration parameter, β i is the regression coefficient, ∈ is the error term, β0 is the reference value, β1,β2,…,β n are the regression coefficients of the calibration parameters, P1, P2, …, P n are the various calibration parameters; Step 3: Based on the regression analysis results, the calibration parameters are adjusted using the gradient descent optimization algorithm. The formula is: in, and are the new and old calibration parameter values, respectively, and α is the learning rate; Optimization result feedback unit: Feedback the optimization results to the dynamic calibration control module and the environmental adaptation module, including updating the calibration parameters to adjust the position and angle of the robotic arm, and adjusting the light source intensity and imaging parameters of the multispectral imaging module.

7. The electronic atomizer chip visual recognition precision calibration system according to claim 1, characterized in that: The interactive user interface module includes an operation interface unit, a real-time feedback display unit, a parameter input unit, a system monitoring unit and a setting adjustment unit; wherein: Operation interface unit: used to provide a user operation interface, allowing the user to input calibration parameters, start or stop the calibration process, and select the calibration mode; Real-time feedback display unit: used to display real-time data and feedback information during the calibration process, including the current chip position, angle, calibration accuracy, and detected defect information; Parameter input unit: supports users to input calibration parameters, including target position x, y, rotation angle θ; System monitoring unit: used to monitor the operating status of the system, including the motion status of the robot arm, the working status of the servo motor, and the parameter adjustment status of the environmental adaptive module; Setting adjustment unit: used to adjust system settings according to user operations and feedback information, including fine-tuning of calibration parameters, response speed adjustment of servo motors, and parameter adjustment of environmental adaptation modules.

Images