Intelligent control method and system for full-automatic plate unloading

Abstract

The invention discloses an intelligent control method and system for full-automatic board unloading, and relates to the technical field of intelligent manufacturing, and the method comprises the steps that a mechanical arm executes grabbing operation according to optimal grabbing parameters, analyzes the grabbing stability of a PCB in real time, and dynamically adjusts the grabbing force and angle of the mechanical arm according to the grabbing stability of the PCB; based on the adjusted grabbing force and angle of the mechanical arm, the plate unloading speed is initialized through a self-adaptive control algorithm; and carrying out board unloading operation according to the initialized board unloading speed, monitoring the actual movement track of the PCB in real time, carrying out board unloading risk assessment, and dynamically adjusting the board unloading speed according to an assessment result. The grabbing stability is analyzed in real time through an energy balance equation, the grabbing force and angle are dynamically adjusted, and the stability of the grabbing process is ensured; the plate unloading speed is initialized in combination with a self-adaptive control algorithm, and the plate unloading speed is dynamically adjusted by adopting nonlinear mapping and a dynamic adjustment mechanism, so that the stability and the safety of the plate unloading process are ensured.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and in particular to an intelligent control method and system for fully automatic panel unloading. Background Art

[0002] With the rapid development of industrial automation and intelligent manufacturing, traditional unloading operations mainly rely on manual or semi-automatic equipment, which has problems such as low efficiency, poor precision, and easy damage to PCB boards. In recent years, the fully automatic unloading system based on machine vision and robot control technology has gradually become a research hotspot. In the prior art, machine vision technology is widely used in the positioning and detection of PCB boards, while robot control technology is used to achieve precise grasping and unloading operations. However, the existing methods still have significant deficiencies in grasping parameter prediction, grasping stability analysis, and unloading speed control.

[0003] The main deficiencies of the existing technology are: first, the prediction of the grasping parameters lacks deep integration of the geometric characteristics of the PCB board and time series data, resulting in insufficient prediction accuracy and inability to adapt to dynamic changes in complex production environments; second, the control of the board unloading speed lacks a dynamic adjustment mechanism based on real-time monitoring and risk assessment, making it difficult to ensure the smoothness and safety of the board unloading process. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent control method for fully automatic unloading of plates to solve the problems of insufficient prediction accuracy of grasping parameters and poor stability of the unloading process.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides an intelligent control method for fully automatic board unloading, which includes collecting visual data of a PCB board and extracting geometric features, using a timing fusion mechanism based on the geometric features to predict optimal grasping parameters of a robotic arm, wherein the visual data includes position coordinates, angle information, contour shape, surface texture, and edge data of the PCB board; the robotic arm performs a grasping operation according to the optimal grasping parameters, and analyzes the grasping stability of the PCB board in real time, and dynamically adjusts the grasping strength and angle of the robotic arm according to the grasping stability of the PCB board; based on the adjusted grasping strength and angle of the robotic arm, an adaptive control algorithm is used to initialize the board unloading speed; the board unloading operation is performed according to the initialized board unloading speed, the actual motion trajectory of the PCB board is monitored in real time, and a board unloading risk assessment is performed, and the board unloading speed is dynamically adjusted according to the assessment result.

[0008] As a preferred solution of the intelligent control method for fully automatic unloading of the plate of the present invention, the specific steps of extracting geometric features are as follows:

[0009] Preprocess the visual data of PCB boards through denoising, contrast enhancement, edge sharpening and image segmentation;

[0010] Based on the preprocessed visual data of the PCB board, the contour detection algorithm, Hough transform, GLCM and edge detection algorithm are used to extract the contour features, angle features, surface texture features and edge features of the PCB board.

[0011] As a preferred solution of the intelligent control method for fully automatic unloading of the plate of the present invention, wherein: the optimal grasping parameters of the robot arm are predicted by using a time series fusion mechanism based on geometric features, and the specific steps are as follows:

[0012] Organize geometric features according to time series and construct time series data;

[0013] Based on the time series data, a bidirectional long short-term memory network is constructed through two LSTM units, and the timing features of the PCB board are extracted through the bidirectional long short-term memory network;

[0014] The geometric features and temporal features are temporally fused through a dynamic interaction mechanism based on graph neural networks to generate geometric temporal features.

[0015] Based on the geometric timing features, nonlinear transformation and enhancement are performed through hyperbolic orthogonal function and exponential function respectively, and Sigmoid function is used for mapping to generate the optimal grasping parameter y of the robot arm.

[0016] As a preferred solution of the intelligent control method for fully automatic unloading of the board of the present invention, wherein: the mechanical arm performs the grasping operation according to the optimal grasping parameters and analyzes the grasping stability of the PCB board in real time. The specific steps are as follows:

[0017] When the robot grabs the PCB board according to the optimal grabbing parameters, the status data of the PCB board is collected;

[0018] Based on the force data of the PCB board, the energy balance equation is established by analyzing the dynamic energy interaction during the grasping process, and the grasping stability index S(t) of the PCB board is calculated;

[0019] Based on historical crawling data, the stability standard S1 is defined through statistical analysis;

[0020] When S(t)≥S1, the robot grasping process is considered to be stable, otherwise it is considered to be unstable.

[0021] As a preferred solution of the intelligent control method for fully automatic unloading of the board of the present invention, wherein: the grasping force and angle of the robot arm are dynamically adjusted according to the grasping stability of the PCB board, and the specific steps are as follows:

[0022] Based on the deviation of the grasping stability index S(t) and the stability standard S1, the grasping force is dynamically adjusted according to the energy balance principle; ˙

[0023] The friction loss E(t) between the robot arm and the PCB board and the vibration speed b(t) of the PCB board are used as adjustment factors to dynamically adjust the grasping angle.

[0024] As a preferred solution of the intelligent control method for fully automatic unloading of the board of the present invention, wherein: based on the adjusted gripping force and angle of the robot arm, the unloading speed is initialized using an adaptive control algorithm, and the specific steps are as follows:

[0025] Formulate adaptive control algorithms based on MRAC and RL learning mechanisms;

[0026] The adjusted gripping force and angle of the robot arm are used as adjustment factors of the adaptive control algorithm;

[0027] The adaptive control algorithm uses MRAC to preliminarily calculate the lowering speed and optimizes it through the RL learning mechanism to generate the initial lowering speed V.

[0028] As a preferred solution of the intelligent control method for fully automatic board unloading of the present invention, the board unloading operation is performed according to the initialization board unloading speed, the motion trajectory of the PCB board is monitored in real time, and the board unloading risk assessment is performed, and the board unloading speed is dynamically adjusted according to the assessment result. The specific steps are as follows:

[0029] During the board removal process, the actual motion trajectory of the PCB board is collected, and the deviation ΔH between the actual motion trajectory and the expected motion trajectory is calculated;

[0030] Based on the statistical analysis of the historical PCB board motion trajectory deviation, the deviation threshold H1 is defined;

[0031] When ΔH≥H1, there is a risk in the board removal process, otherwise, there is no risk in the board removal process;

[0032] According to the risk assessment results, nonlinear mapping and dynamic adjustment mechanism are used to dynamically adjust the board unloading speed.

[0033] In a second aspect, the present invention provides an intelligent control system for fully automatic board unloading, including a grasping parameter prediction module, a robotic arm adjustment module, a board unloading speed initialization module and a board unloading speed adjustment module; the grasping parameter prediction module is used to collect visual data of the PCB board and extract geometric features, and use a timing fusion mechanism based on the geometric features to predict the optimal grasping parameters of the robotic arm; the robotic arm adjustment module is used for the robotic arm to perform a grasping operation according to the optimal grasping parameters, and to analyze the grasping stability of the PCB board in real time, and to dynamically adjust the grasping force and angle of the robotic arm according to the grasping stability of the PCB board; the board unloading speed initialization module is used to initialize the board unloading speed based on the adjusted grasping force and angle of the robotic arm using an adaptive control algorithm; the board unloading speed adjustment module is used to perform the board unloading operation according to the initialized board unloading speed, monitor the actual motion trajectory of the PCB board in real time, and perform a board unloading risk assessment, and dynamically adjust the board unloading speed according to the assessment result.

[0034] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the intelligent control method for fully automatic board unloading as described in the first aspect of the present invention is implemented.

[0035] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the intelligent control method for fully automatic plate unloading as described in the first aspect of the present invention is implemented.

[0036] The beneficial effects of the present invention are as follows: by collecting visual data of the PCB board and extracting geometric features, the optimal grasping parameters are predicted using a timing fusion mechanism to ensure the accuracy and adaptability of the grasping operation; the grasping operation is performed based on the optimal grasping parameters, and the grasping stability is analyzed in real time through an energy balance equation, and the grasping force and angle are dynamically adjusted to ensure the stability of the grasping process; the board unloading speed is initialized in combination with an adaptive control algorithm, and the board unloading speed is dynamically adjusted using a nonlinear mapping and dynamic adjustment mechanism to ensure the stability and safety of the board unloading process. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0038] Figure 1 This is a flow chart of the intelligent control method for fully automatic plate unloading in Example 1.

[0039] Figure 2Schematic diagram of the intelligent control system for fully automatic unloading of plates in Example 1.

[0040] Figure 3 This is a flow chart of the timing fusion mechanism in the intelligent control method for fully automatic board unloading in Example 1.

[0041] Figure 4 This is a flow chart of the dynamic adjustment of the plate lowering speed in the intelligent control method for fully automatic plate lowering in Example 1. DETAILED DESCRIPTION

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0043] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0044] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0045] Example 1, reference Figure 1 to Figure 4 , which is the first embodiment of the present invention, provides an intelligent control method for fully automatic unloading of the plate, comprising the following steps:

[0046] S1. Collect visual data of PCB boards and extract geometric features. Use the timing fusion mechanism based on the geometric features to predict the optimal grasping parameters of the robot arm.

[0047] The visual data includes the position coordinates, angle information, contour shape, surface texture and edge data of the PCB board.

[0048] It should be noted that a camera installed above the working area is used to capture real-time images of the PCB board to ensure that the entire operating area is covered to obtain comprehensive information. The position coordinates and angle information are used with the help of the feature point detection algorithm in the image processing technology to identify the key points of the PCB board and calculate its position and angle relationship relative to the preset reference point. The contour shape is determined by analyzing the area where the brightness in the image changes significantly through the edge detection algorithm, thereby accurately outlining the shape of the PCB board. The surface texture is extracted from the image by contrast enhancement and texture analysis algorithms to identify surface characteristics and possible defects. The edge data is processed by a special edge detection algorithm to accurately locate the edge of the PCB board.

[0049] Preprocess the visual data of PCB boards through denoising, contrast enhancement, edge sharpening and image segmentation;

[0050] Furthermore, the visual data of the PCB board is optimized through denoising, and a Gaussian filter is used to remove random noise in the image, ensuring that the value of each pixel in the image is affected by the surrounding pixels, thereby smoothing the image and reducing pseudo features caused by noise. For example, when detecting tiny defects on the PCB board, the denoising step can effectively reduce the interference introduced by sensors or environmental factors.

[0051] The contrast enhancement process uses histogram equalization technology to adjust the grayscale distribution of the image to expand the brightness difference between different areas in the image and improve the overall clarity of the visual data. Specifically, for the situation where the surface of the PCB board may have different shades of color, contrast enhancement can make the surface texture more obvious, facilitating the accurate identification of various features, such as solder joints or component logos.

[0052] Edge sharpening operation uses the Laplacian operator to highlight the boundary information in the image, enhancing the display effect of the outline and details of the PCB board. For example, when identifying the fine edge of the PCB board, edge sharpening can increase the contrast between the edge and the background, making the originally blurred edge clear and sharp, which is conducive to accurate measurement and positioning.

[0053] Image segmentation divides the visual data of the PCB into multiple meaningful parts or regions based on a specific algorithm for separate analysis. For example, the threshold segmentation method can be used to divide the image into the target area (such as the PCB itself) and the background area based on a preset grayscale threshold.

[0054] Based on the preprocessed visual data of the PCB board, the contour detection algorithm, Hough transform, GLCM and edge detection algorithm are used to extract the contour features, angle features, surface texture features and edge features of the PCB board.

[0055] Organize geometric features according to time series and construct time series data;

[0056] Furthermore, it is necessary to first annotate and record the geometric features in the continuously collected multi-frame PCB board visual data. The geometric features in each frame of the image, including position coordinates, angle information, contour shape, surface texture and edge data, are assigned corresponding timestamps to reflect the changes of these features over time. For example, in an automatic board unloading operation cycle, from the time the robot arm approaches the PCB board until the completion of the grabbing, an image is collected at a fixed time interval (such as every 10 milliseconds) and all geometric features at that moment are extracted. Then, these geometric features are connected in series according to the time sequence to form one or more time series data about the evolution of specific geometric features over time.

[0057] Based on the time series data, a bidirectional long short-term memory network is constructed through two LSTM units, and the timing features of the PCB board are extracted through the bidirectional long short-term memory network;

[0058] Furthermore, the first LSTM unit is used to forward propagate the time series data in chronological order to capture the temporal dependencies of past moments, and the second LSTM unit is used to back propagate the time series data in reverse chronological order to capture the temporal dependencies of future moments.

[0059] The specific process includes: first, use the first LSTM unit to forward propagate the data in chronological order, which means starting from the earliest time point and gradually moving forward to the latest time point, so as to capture the temporal dependencies of past moments. For example, in the process of analyzing the movement of the PCB board from the initial position to the grasped point, the first LSTM unit will process the geometric features corresponding to each frame of the image in turn, and learn the trend of position coordinates and angle information changing over time. At the same time, the second LSTM unit processes the same time series data in reverse chronological order, that is, backtracking from the latest time point to the earliest time point, so as to capture the temporal dependencies of future moments. Taking the same PCB board movement process as an example, the second LSTM unit can identify the impact of subtle changes near the completion stage of grasping on future operation steps.

[0060] By concatenating the hidden states of the first LSTM unit and the second LSTM unit at each time step, a bidirectional long short-term memory network is constructed, and the back propagation algorithm and gradient descent method are used for training based on historical time series data;

[0061] The specific process includes: constructing a bidirectional long short-term memory network (Bi-LSTM) by splicing the hidden states of the first LSTM unit and the second LSTM unit at each time step, so as to fully capture the forward and backward dependencies in the time series data of geometric features. Specifically, when processing the process of the PCB board moving from the initial position to the grasped point, for the geometric features corresponding to each frame image, the first LSTM unit extracts the information of the past moment from the forward propagation, while the second LSTM unit extracts the information of the future moment from the backward propagation. At the end of each time step, the hidden states generated by the two LSTM units are spliced ​​together to form a vector that comprehensively represents the information before and after the time step. Based on these spliced ​​hidden states and historical time series data, the back propagation algorithm is used to calculate the gradient of the loss function relative to the network weight, and the gradient descent method is used to update the network weight to minimize the prediction error. For example, in the process of optimizing the grasping parameters of the robot arm, through repeated iterative training processes, the Bi-LSTM network can more accurately predict the optimal grasping parameters suitable for different situations, thereby improving the accuracy and efficiency of the operation.

[0062] The geometric features and time series features are fused in time series through a dynamic interaction mechanism based on graph neural networks to generate geometric time series features;

[0063] Specifically, the geometric features and temporal features are constructed as a graph structure, where each node represents a specific geometric feature or temporal feature, and the edge represents the dependency between the geometric features and the temporal features. For example, when processing the grasping operation of a PCB board, geometric features such as position coordinates, angle information, etc. together with the temporal features extracted from the bidirectional long short-term memory network (Bi-LSTM) constitute the nodes of the graph, and the edges between the nodes reflect how the geometric features and the temporal features affect each other, such as how the position change affects the choice of the grasping angle. Using the graph neural network (GNN), it is possible to capture the complex interactions between these features, and dynamically update the feature representation of each node in each iteration through the graph convolution operation. Specifically, the graph convolution operation allows the node to aggregate the information of its neighboring nodes, so as to better understand the interaction between local features and the long-range dependencies in the overall graph structure. This process enables the combination of each geometric feature and the temporal feature to be continuously optimized, and finally generates a geometric temporal feature that integrates temporal and spatial information.

[0064] Based on the geometric timing characteristics, nonlinear transformation and enhancement are performed through hyperbolic orthogonal function and exponential function respectively, and Sigmoid function is used for mapping to generate the optimal grasping parameter y of the robot arm, which is expressed as:

[0065]

[0066] Among them, y is the optimal capture parameter, n is the total number of geometric time series features, i is the index variable of geometric time series features, σ is Sigmoid, α i is the weight coefficient of the i-th geometric time series feature, W 1 is the weight matrix of the nonlinear transformation (through the tanh function) of the geometric time series features, W 2 is the weight matrix of geometric time series features enhanced (through exp function), b 1 is the bias term of the geometric time series feature after nonlinear transformation, b 2 is the bias term of the enhanced geometric timing feature, h i is the i-th geometric timing feature;

[0067] It should be noted that after constructing the geometric timing features, in order to determine the influence of each feature on generating the optimal grasping parameter y of the robot arm, α is automatically learned through the optimization algorithm in the training process. i This process takes into account the relationship between several time-series features and the final grasping effect, ensuring that features that are critical to the grasping operation receive higher weights. i The value range of is any real number in the initial stage, but after α is converted to i Convert to probability form so that all α i The value of is between 0 and 1 and satisfies conditions.

[0068] The optimal grasping parameters include grasping force, grasping angle and grasping position.

[0069] S2. The robot arm performs the grasping operation according to the optimal grasping parameters, and analyzes the grasping stability of the PCB board in real time. According to the grasping stability of the PCB board, the grasping force and angle of the robot arm are dynamically adjusted.

[0070] When the robot grabs the PCB board according to the optimal grabbing parameters, the status data of the PCB board is collected;

[0071] The status data of the PCB board includes gripping force, displacement, torque, angle change, linear velocity, angular velocity, elastic deformation and vibration speed;

[0072] Based on the force data of the PCB board, the energy balance equation is established by analyzing the dynamic energy interaction during the grasping process, and the grasping stability index S(t) of the PCB board is calculated. The expression is:

[0073]

[0074] Among them, S(t) represents the grasping stability index at time t, F(t) represents the grasping force applied by the robot at time t, d(t) represents the displacement of the robot at time t, M(t) represents the torque applied by the robot at time t, θ(t) represents the angular change of the robot at time t, E(t) represents the friction loss at time t, q represents the mass of the PCB board, g(t) represents the linear velocity of the PCB board at time t, I represents the moment of inertia of the PCB board, ω(t) represents the angular velocity of the PCB board at time t, and k represents the ˙ Elastic coefficient, x(t) represents the elastic variable of the PCB board at time t, c represents the damping coefficient of the PCB board, and b(t) represents the vibration speed of the PCB board at time t;

[0075] It should be explained that the process of establishing the energy balance equation for dynamic energy interaction is as follows: First, considering that the force and torque applied by the robot arm on the PCB board produce linear displacement and angular change respectively, according to the principles of physics, these actions can be converted into work done on the PCB board, that is, F(t)·d(t) represents the linear work caused by the grasping force, and M(t)·θ(t) represents the rotational work caused by the torque. These two parts together constitute the total energy input to the PCB board from the outside. However, in actual operation, due to the existence of factors such as friction, part of the energy will be lost, and this part of the lost energy will be deducted from the total energy in the form of friction loss E(t). On the other hand, in order to evaluate whether the PCB board is stable during this process, it is necessary to consider its kinetic energy state, including linear kinetic energy. Rotational energy Elastic potential energy and the kinetic energy caused by vibration Integrating all the above factors, we can form an energy balance equation, where the numerator is the net energy of external input (i.e. (F(t)·d(t)+M(t)·θ(t))-E(t)), and the denominator is the total kinetic energy of the PCB. The grasping stability index S(t) is calculated from this ratio.

[0076] It should also be noted that the mass q of the CB board is obtained by measuring the actual weight of the PCB board, the moment of inertia I of the PCB board is determined by calculating the geometric shape and mass distribution of the PCB board, the elastic coefficient k of the PCB board is obtained by experimentally measuring the deformation and recovery characteristics of the PCB board when subjected to force, and the damping coefficient c of the PCB board is determined by analyzing the energy dissipation characteristics of the PCB board during vibration. The linear velocity g(t), angular velocity ω(t), elastic variable x(t) and vibration velocity b(t) of the PCB board are obtained by real-time acquisition of the motion state and vibration data of the PCB board during the grasping process by sensors.

[0077] Based on historical crawling data, the stability standard S1 is defined through statistical analysis;

[0078] When S(t)≥S1, the robot grasping process is considered to be stable, otherwise it is considered to be unstable.

[0079] Based on the deviation between the grasping stability index S(t) and the stability standard S1, the grasping force is dynamically adjusted according to the energy balance principle. The expression is:

[0080] F'(t)=F(t)+ΔF·(S1-S(t));

[0081] Where F′(t) is the adjusted grasping force, and ΔF represents the adjustment step of the grasping force;

[0082] It should be noted that the process of dynamically adjusting the grasping force based on the deviation between the grasping stability index and the stability standard and in accordance with the energy balance principle relies on the energy evaluation and optimization of the current grasping state. When the grasping operation is detected to be insufficiently stable, it means that the currently applied force fails to effectively maintain the PCB board in the expected stable state. At this time, it is necessary to compensate for this instability by increasing the grasping force to ensure that the robot arm can firmly and safely grasp the PCB board. On the contrary, if the grasping is too stable or even exceeds the necessary range, the grasping force is appropriately reduced to prevent potential damage caused by excessive constraints. This process not only takes into account the forces and torques acting directly on the PCB board, but also combines the influence of factors such as friction loss and vibration speed. This information is integrated to make real-time adjustments, aiming to minimize energy consumption while ensuring the safety and stability of the grasping process. ˙

[0083] The friction loss E(t) between the robot arm and the PCB board and the vibration speed b(t) of the PCB board are used as adjustment factors to dynamically adjust the gripping angle. The expression is:

[0084]

[0085] Among them, θ′(t) is the adjusted grasping angle, and Δθ represents the adjustment step of the grasping angle.

[0086] It should be noted that the principle of optimizing the grasping process by dynamically adjusting the grasping angle based on the friction loss between the robot arm and the PCB board and the vibration speed of the PCB board is to accurately control the angle when the robot arm contacts the PCB board to adapt to different operating conditions and ensure the best grasping effect. During the grasping process, the friction loss reflects the loss of energy between the contact surfaces, while the vibration speed indicates the dynamic behavior of the PCB board at the moment of grasping. Using these two factors as adjustment factors, the energy efficiency and stability under the current grasping configuration can be evaluated in real time, and the grasping angle can be adjusted accordingly. When the friction loss is large or the vibration speed exceeds expectations, it indicates that the existing grasping angle may not be suitable for the current conditions, and corresponding adjustments are required to reduce energy loss and suppress unnecessary vibration, so as to ensure that the grasping process is both stable and efficient. This adaptive adjustment mechanism based on actual operating parameters can effectively improve the accuracy and safety of the robot arm's grasping of PCB boards, and also help to extend the service life of the equipment.

[0087] S3. Based on the adjusted gripping force and angle of the robot arm, the unloading speed is initialized using an adaptive control algorithm.

[0088] Formulate adaptive control algorithms based on MRAC and RL learning mechanisms;

[0089] Furthermore, relying on the pre-set ideal behavior standard, the difference between the actual operation state and the standard is evaluated in real time, and the dynamic adjustment strategy is used to narrow the gap between the two to ensure efficient performance under known conditions. At the same time, with the help of RL learning mechanism, by continuously learning and accumulating experience from a large number of operation examples, it is possible to explore and implement the optimal operation strategy in an uncertain or changing working environment. This method can identify different working conditions, continuously optimize the decision-making process based on feedback information, and quickly respond to and adapt to new situations. The final intelligent adaptive control algorithm combines the advantages of these two methods: on the one hand, it handles foreseeable changes through precise dynamic adjustment; on the other hand, it uses learning mechanisms to deal with unknown or unpredictable situations. This algorithm can automatically adjust key parameters such as gripping force and board unloading speed, which not only ensures the stability and safety of operation, but also significantly improves overall efficiency and flexibility.

[0090] The adjusted gripping force and angle of the robot arm are used as adjustment factors of the adaptive control algorithm;

[0091] It should be noted that the gripping force is adjusted according to the deviation between the gripping stability index and the preset standard to ensure that sufficient gripping stability can be maintained under different conditions while avoiding damage caused by excessive force. At the same time, factors such as friction loss and vibration speed are used to further fine-tune the gripping angle to compensate for any factors that may cause instability. This method enables the adaptive control algorithm to make precise adjustments based on feedback from actual operations, which not only improves the stability and safety of the gripping process, but also enhances the adaptability to various working conditions.

[0092] The adaptive control algorithm uses MRAC to preliminarily calculate the lowering speed and optimizes it through the RL learning mechanism to generate the initial lowering speed V, which is expressed as:

[0093]

[0094] Where V is the initial board speed, g(F′(t),θ′(t)) is the initial estimate of the board speed mapped by the nonlinear mapping function, η is the weight coefficient of the RL learning mechanism, Q(s,a) is the state-action value function in the RL learning mechanism, represents the expected cumulative benefit of taking action a in state s.

[0095] It should be noted that η, as the weight coefficient of the RL learning mechanism, is defined based on the need to balance reinforcement learning feedback with traditional control strategies. The value of η determines the degree of influence of the experience (i.e., the state-action value function) obtained from the reinforcement learning mechanism on the final decision. Specifically, η is used to adjust the emphasis of the reinforcement learning part on future expected returns, so as to gradually introduce the optimization strategy obtained through learning while following the preset behavioral standards. The value range of η is a real number between 0 and 1, where a value close to 0 indicates a preference for relying on the initially set behavioral standards and precise dynamic adjustment strategies, while a value close to 1 indicates a high emphasis on the experience accumulated through reinforcement learning and the optimal strategy explored.

[0096] The expression of the nonlinear mapping function is:

[0097]

[0098] Among them, k 1 represents the scaling factor, which is used to adjust the amplitude of the mapping result, and r is the parameter that controls the mapping width;

[0099] The expression of the state-action value function is:

[0100]

[0101] Among them, s is the current state, a is the current action, m is the time step, r mis the immediate reward at time step m, γ is the discount factor, R represents the expected value, and s 0 represents the initial state, a 0 Indicates the initial action.

[0102] S4. Execute the board unloading operation according to the initialized board unloading speed, monitor the actual movement trajectory of the PCB board in real time, and conduct board unloading risk assessment, and dynamically adjust the board unloading speed according to the assessment results.

[0103] During the board removal process, the actual motion trajectory of the PCB board is collected, and the deviation ΔH between the actual motion trajectory and the expected motion trajectory is calculated;

[0104] Specifically, the expected motion trajectory is defined based on a series of coordinate points that the PCB needs to pass through from the starting position to the target position. These coordinate points take into account factors such as the robot's motion capability, speed planning, and obstacles in the operating environment. The expected motion trajectory is a series of continuous time functions, and each time point corresponds to the spatial position where the PCB should be located. For example, in three-dimensional space, the expected motion trajectory can be expressed in the form of three-dimensional coordinates. The actual motion trajectory is determined by real-time collected data, including the actual position of the robot at each moment. The deviation ΔH between the actual motion trajectory and the expected motion trajectory is calculated, that is, the difference between the actual position and the expected position at each time point is compared to evaluate the accuracy and stability of the entire unloading process.

[0105] Based on the statistical analysis of the historical PCB board motion trajectory deviation, the deviation threshold H1 is defined;

[0106] When ΔH≥H1, there is a risk in the board removal process, otherwise, there is no risk in the board removal process;

[0107] According to the risk assessment results, nonlinear mapping and dynamic adjustment mechanism are used to dynamically adjust the board speed. The expression is:

[0108]

[0109] Among them, V′(t) is the lower plate speed after dynamic adjustment, V(t) is the lower plate speed at the current time t, ΔV represents the adjustment step of the lower plate speed, u is the scaling parameter of the stability deviation, and p is the scaling parameter of the vibration deviation.

[0110] It should be noted that the hyperbolic tangent function tanh is used to convert this stability deviation into an adjustment factor that can smoothly reflect the change process from unstable to stable state. At the same time, considering the vibration problem that may occur during the operation of the robot arm, the exponential function exp is used to quantify the vibration deviation caused by the difference between the actual vibration speed b(t) and the friction loss E(t), and the square term is used to further emphasize the impact of the larger deviation. These two adjustment factors are combined to dynamically calculate the adjustment amount of the lower plate speed. In this way, when a larger stability or vibration deviation is detected, the adjusted lower plate speed V′(t) will be reduced accordingly to increase the stability of the operation; on the contrary, when the deviation is small, the lower plate speed can be appropriately increased to optimize the efficiency. The whole process realizes fine control of the lower plate speed through nonlinear mapping and dynamic adjustment mechanism to ensure that the efficiency is maximized while ensuring the safety of operation.

[0111] The present embodiment also provides an intelligent control system for fully automatic board unloading, including: a grasping parameter prediction module, a mechanical arm adjustment module, a board unloading speed initialization module and a board unloading speed adjustment module; the grasping parameter prediction module is used to collect visual data of the PCB board and extract geometric features, and use a timing fusion mechanism based on the geometric features to predict the optimal grasping parameters of the mechanical arm; the mechanical arm adjustment module is used for the mechanical arm to perform a grasping operation according to the optimal grasping parameters, and to analyze the grasping stability of the PCB board in real time, and to dynamically adjust the grasping force and angle of the mechanical arm according to the grasping stability of the PCB board; the board unloading speed initialization module is used to initialize the board unloading speed based on the adjusted grasping force and angle of the mechanical arm using an adaptive control algorithm; the board unloading speed adjustment module is used to perform the board unloading operation according to the initialized board unloading speed, monitor the actual motion trajectory of the PCB board in real time, and perform a board unloading risk assessment, and dynamically adjust the board unloading speed according to the assessment result.

[0112] This embodiment also provides a computer device, which is suitable for the intelligent control method of fully automatic board unloading, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the intelligent control method of fully automatic board unloading proposed in the above embodiment.

[0113] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0114] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by the processor, the intelligent control method for realizing fully automatic board unloading as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0115] In summary, the present invention collects visual data of the PCB board and extracts geometric features, uses a timing fusion mechanism to predict optimal grasping parameters, and ensures the accuracy and adaptability of the grasping operation; performs grasping operations based on the optimal grasping parameters, and analyzes grasping stability in real time through an energy balance equation, dynamically adjusts grasping force and angle, and ensures the stability of the grasping process; initializes the board unloading speed in combination with an adaptive control algorithm, and dynamically adjusts the board unloading speed using a nonlinear mapping and dynamic adjustment mechanism, and ensures the stability and safety of the board unloading process.

[0116] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent control method for fully automatic unloading of boards, characterized in that: include, Collect visual data of the PCB board and extract geometric features, and use a time fusion mechanism based on the geometric features to predict the optimal grasping parameters of the robot arm. The visual data includes the position coordinates, angle information, contour shape, surface texture and edge data of the PCB board. The robot arm performs the grasping operation according to the optimal grasping parameters, and analyzes the grasping stability of the PCB board in real time. According to the grasping stability of the PCB board, the grasping force and angle of the robot arm are dynamically adjusted; Based on the adjusted gripping force and angle of the robot arm, the unloading speed is initialized using an adaptive control algorithm; The unloading operation is performed according to the initialized unloading speed, the actual movement trajectory of the PCB board is monitored in real time, and the unloading risk assessment is performed, and the unloading speed is dynamically adjusted according to the assessment results.

2. The intelligent control method for fully automatic unloading of plates as claimed in claim 1, characterized in that: The specific steps of extracting geometric features are as follows: Preprocess the visual data of PCB boards through denoising, contrast enhancement, edge sharpening and image segmentation; Based on the preprocessed visual data of the PCB board, the contour detection algorithm, Hough transform, GLCM and edge detection algorithm are used to extract the contour features, angle features, surface texture features and edge features of the PCB board.

3. The intelligent control method for fully automatic unloading of plates as claimed in claim 2, characterized in that: The specific steps of using the time series fusion mechanism based on geometric features to predict the optimal grasping parameters of the robot arm are as follows: Organize geometric features according to time series and construct time series data; Based on the time series data, a bidirectional long short-term memory network is constructed through two LSTM units, and the timing features of the PCB board are extracted through the bidirectional long short-term memory network; The geometric features and time series features are fused in time series through a dynamic interaction mechanism based on graph neural networks to generate geometric time series features; Based on the geometric timing features, nonlinear transformation and enhancement are performed through hyperbolic orthogonal function and exponential function respectively, and Sigmoid function is used for mapping to generate the optimal grasping parameter y of the robot arm.

4. The intelligent control method for fully automatic unloading of plates as claimed in claim 3, characterized in that: The robot arm performs the grasping operation according to the optimal grasping parameters and analyzes the grasping stability of the PCB board in real time. The specific steps are as follows: When the robot grabs the PCB board according to the optimal grabbing parameters, the status data of the PCB board is collected; Based on the force data of the PCB board, the energy balance equation is established by analyzing the dynamic energy interaction during the grasping process, and the grasping stability index S(t) of the PCB board is calculated; Based on historical crawling data, the stability standard S1 is defined through statistical analysis; When S(t)≥S1, the robot grasping process is considered to be stable, otherwise it is considered to be unstable.

5. The intelligent control method for fully automatic unloading of plates as claimed in claim 4, characterized in that: According to the grasping stability of the PCB board, the grasping force and angle of the robot arm are dynamically adjusted. The specific steps are as follows: Based on the deviation of the grasping stability index S(t) and the stability standard S1, the grasping force is dynamically adjusted according to the energy balance principle; ˙ The friction loss E(t) between the robot arm and the PCB board and the vibration speed b(t) of the PCB board are used as adjustment factors to dynamically adjust the grasping angle.

6. The intelligent control method for fully automatic unloading of plates as claimed in claim 5, characterized in that: Based on the adjusted gripping force and angle of the robot arm, the adaptive control algorithm is used to initialize the board lowering speed. The specific steps are as follows: Formulate adaptive control algorithms based on MRAC and RL learning mechanisms; The adjusted gripping force and angle of the robot arm are used as adjustment factors of the adaptive control algorithm; The adaptive control algorithm uses MRAC to preliminarily calculate the lowering speed and optimizes it through the RL learning mechanism to generate the initial lowering speed V.

7. The intelligent control method for fully automatic unloading of plates as claimed in claim 6, characterized in that: The board unloading operation is performed according to the initialization board unloading speed, the movement trajectory of the PCB board is monitored in real time, and the board unloading risk assessment is performed, and the board unloading speed is dynamically adjusted according to the assessment result. The specific steps are as follows: During the board removal process, the actual motion trajectory of the PCB board is collected, and the deviation ΔH between the actual motion trajectory and the expected motion trajectory is calculated; Based on the statistical analysis of the historical PCB board motion trajectory deviation, the deviation threshold H1 is defined; When ΔH≥H1, there is a risk in the board removal process, otherwise, there is no risk in the board removal process; According to the risk assessment results, nonlinear mapping and dynamic adjustment mechanism are used to dynamically adjust the board unloading speed.

8. An intelligent control system for fully automatic unloading of panels, based on the intelligent control method for fully automatic unloading of panels as claimed in any one of claims 1 to 7, characterized in that: It includes a grasping parameter prediction module, a robotic arm adjustment module, a lower plate speed initialization module and a lower plate speed adjustment module; The grasping parameter prediction module is used to collect visual data of PCB boards and extract geometric features. Based on the geometric features, the timing fusion mechanism is used to predict the optimal grasping parameters of the robot arm. The robot arm adjustment module is used for the robot arm to perform the grasping operation according to the optimal grasping parameters, and to analyze the grasping stability of the PCB board in real time, and to dynamically adjust the grasping force and angle of the robot arm according to the grasping stability of the PCB board; The board lowering speed initialization module is used to initialize the board lowering speed using an adaptive control algorithm based on the adjusted gripping force and angle of the robot arm; The board unloading speed adjustment module is used to perform the board unloading operation according to the initialized board unloading speed, monitor the actual movement trajectory of the PCB board in real time, and perform board unloading risk assessment, and dynamically adjust the board unloading speed according to the assessment results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the intelligent control method for fully automatic unloading of the plate as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent control method for fully automatic unloading of panels as claimed in any one of claims 1 to 7 are implemented.

Images