Chip mounter mounting precision fluctuation optimization method and system

Abstract

The invention provides a chip mounter mounting precision fluctuation optimization method and system. The method comprises the steps that potential influence variables influencing mounting precision fluctuation are divided into easy-to-adjust variables and difficult-to-adjust variables; obtaining mounting deviation data under different easy-to-adjust variable combinations through experimental design; based on the mounting deviation data, establishing a first quantitative relation model between the easy-to-adjust variable and mounting precision fluctuation; key easy-to-adjust variables influencing the surface mounting precision fluctuation are identified, and an optimal easy-to-adjust variable parameter set enabling the surface mounting precision fluctuation to be minimum is obtained; uniformly applying the optimal variable parameter set easy to adjust on a plurality of chip mounters with the same model; collecting an actually measured value of a variable difficult to adjust of each chip mounter and mounting deviation data corresponding to the actually measured value to form a multi-machine sample library; establishing a second quantitative relation model between the difficult-to-adjust variable and the mounting precision fluctuation by adopting a machine learning algorithm; and identifying one or more key variables difficult to adjust and an optimization tolerance range which should be satisfied by the key variables difficult to adjust.

Description

Technical Field

[0001] This invention relates to the field of surface mount technology, and in particular to a method and system for optimizing the mounting accuracy fluctuation of a pick-and-place machine. Background Technology

[0002] Surface Mount Technology (SMT) is the core of modern electronics manufacturing, with pick-and-place machines being key equipment; fluctuations in placement accuracy directly impact product quality. Existing technologies primarily improve accuracy through motion control, software algorithms, visual inspection, and mechanical structures. For example, positioning accuracy is improved by using linear motors and optimizing servo control algorithms; efficiency is enhanced through path planning algorithms; position calibration is performed using high-resolution cameras and image processing algorithms; and mechanical stability is ensured through high-rigidity materials and precision assembly.

[0003] Although existing pick-and-place machines have made improvements and progress in motion control, vision inspection, mechanical design, and software algorithms, overall, the existing technology still suffers from problems such as insufficient dynamics and intelligence in parameter optimization, lack of real-time closed-loop feedback and predictive capabilities, data fragmentation between subsystems, and mutual constraints between operating speed and accuracy. Specifically, these include: Unclear Coupling of Influencing Factors and Blind Optimization: Existing technologies treat subsystems such as mechanical assembly, motion control, and visual inspection as independent modules for optimization. However, the influencing factors between these systems (such as mechanical assembly tolerances, motor acceleration / deceleration parameters, and visual calibration errors) exhibit complex coupling relationships. Currently, there is a lack of a systematic method to decouple these variables and quantify their respective contribution weights to the final mounting accuracy fluctuations. This leads to parameter optimization relying heavily on manual experience and static calibration, which is not only inefficient but also makes it difficult to find the globally optimal solution, resulting in blind and limited accuracy improvements.

[0004] The lack of a precise quantitative model for the relationship between multi-source data and accuracy fluctuations: Existing technologies struggle to establish a precise quantitative model that encompasses electromechanical assembly data (such as guide rail parallelism and spline shaft runout), motion control parameters (such as acceleration and deceleration parameters for each axis and settling time), and real-time placement process data (such as placement deviations). Because specific electromechanical assembly errors or control parameter variables cannot be quantitatively correlated with the final placement accuracy fluctuations, the system cannot predict accuracy degradation trends, nor can it provide a data-driven scientific basis for setting assembly tolerances for key components and precisely adjusting control parameters.

[0005] The lack of differentiated design for X / Y axis control parameters: Most existing technologies use the same set of control parameters for both the X and Y axes, a fundamental flaw in this "parameter homogenization" design philosophy. It ignores the essential differences between the X and Y axes in terms of mechanical structure characteristics and moment of inertia distribution. These differences mean that the same set of control parameters cannot simultaneously meet the optimal dynamic performance requirements of both axes, instead exacerbating system vibration, prolonging settling time, and ultimately affecting placement efficiency and accuracy. More specifically, existing technologies lack sufficient consideration of the dual-axis coupling effect. During high-speed motion, the X-axis motion dynamically disturbs the Y-axis, and vice versa, and the intensity and characteristics of this mutual influence are asymmetrical along the two axes. Using the same control parameters cannot effectively suppress this asymmetrical coupled vibration, causing the dynamic error of the placement head on the composite motion trajectory to exceed the allowable range. Summary of the Invention

[0006] This invention provides a method and system for optimizing the placement accuracy fluctuation of a pick-and-place machine, which solves the defects of the prior art, such as unclear coupling of influencing factors, lack of accurate quantitative models, and blind optimization process, and realizes scientific, quantitative and source optimization of placement accuracy fluctuation.

[0007] This invention provides a method for optimizing the placement accuracy fluctuation of a pick-and-place machine, comprising: The multiple potential variables affecting the fluctuation of mounting accuracy are divided into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment. On a pick-and-place machine, placement deviation data under different combinations of easily adjustable variables are obtained through experimental design; based on the placement deviation data, a first quantitative relationship model between easily adjustable variables and placement accuracy fluctuation is established; and according to the first quantitative relationship model, key easily adjustable variables affecting placement accuracy fluctuation are identified, and the optimal set of easily adjustable variable parameters that minimizes placement accuracy fluctuation is obtained. On multiple pick-and-place machines of the same model, the optimal set of easily adjustable variable parameters is uniformly applied; the measured values ​​of the difficult-to-adjust variables and their corresponding placement deviation data are collected for each pick-and-place machine to form a multi-machine sample library; based on the multi-machine sample library, a second quantitative relationship model between the difficult-to-adjust variables and the placement accuracy fluctuation is established using a machine learning algorithm; and based on the second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target is derived in reverse for the key difficult-to-adjust variables.

[0008] According to the pick-and-place machine mounting accuracy fluctuation optimization method provided by the present invention, after dividing the variables into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment, the method further includes: Construct a network diagram of fluctuation propagation paths that describes the error propagation relationships among various potential influencing variables; Based on the aforementioned fluctuation propagation path network diagram, and combined with expert knowledge, the potential influencing variables are initially screened to determine the list of variables to enter the optimization process.

[0009] According to the optimization method for placement accuracy fluctuation of a pick-and-place machine provided by the present invention, the establishment of a first quantitative relationship model between easily adjustable variables and placement accuracy fluctuation specifically includes: Experiments were conducted using full-factor or partial-factor experimental design methods, and the first quantitative relationship model was established using response surface methodology or multiple linear regression.

[0010] According to the pick-and-place machine mounting accuracy fluctuation optimization method provided by the present invention, the step of identifying key easily adjustable variables affecting mounting accuracy fluctuation based on the first quantitative relationship model, and obtaining the optimal easily adjustable variable parameter set that minimizes mounting accuracy fluctuation, specifically includes: Analyze the significance of the coefficients of each easily adjustable variable in the first quantitative relationship model; Analyze the factor plots of the main effects or cross effects of each easily adjustable variable; Based on the factor plot, one or more key adjustable variables that have a significant impact on the fluctuation of mounting accuracy are identified; Based on the first quantitative relationship model, with the optimization objective of minimizing the fluctuation of mounting accuracy, the one or more key adjustable variables are synergistically optimized to solve for the optimal set of adjustable variable parameters.

[0011] Specifically, two evaluation indicators were used to identify easily adjustable variables that significantly affect the fluctuation of mounting accuracy. First, the significance of the coefficients corresponding to the easily adjustable variables in the first quantitative relationship model was used. If the coefficients of the variables in the model were significant, then the variables were considered to have a significant impact on the fluctuation of mounting accuracy. Second, factor plots were used. The main effect factor plot showed the fluctuation of mounting accuracy of a single easily adjustable variable as it changed from a low level to a high level. The steeper the straight line, the more significant the impact of the easily adjustable variable on the fluctuation of mounting accuracy. The cross-effect factor plot showed whether there was coupling between two easily adjustable variables. If the two lines in the plot were parallel, it meant that there was no coupling between the variables; otherwise, there was coupling. Subsequently, based on this first quantitative relationship model, a collaborative optimization was performed with the goal of minimizing the comprehensive index of mounting accuracy fluctuation in the X and Y directions. This solved for the optimal selection of easily adjustable variable parameters that minimized the fluctuation of mounting accuracy, thus obtaining the optimized parameter set.

[0012] According to the pick-and-place machine mounting accuracy fluctuation optimization method provided by the present invention, the step of collaboratively optimizing the one or more key easily adjustable variables further includes: In view of the differences in mechanical structure, motion inertia and coupling effect between the X and Y axes of the pick-and-place machine, based on the first quantitative relationship model, collaborative optimization is carried out with the comprehensive index of minimizing the placement accuracy fluctuation in the X and Y directions as the optimization objective, so as to obtain an optimal set of easily adjustable variable parameters that includes axially differentiated motion control parameters with different parameter values ​​for the X and Y axes.

[0013] Specifically, considering the differences in mechanical structure, motion inertia, and coupling effects between the X and Y axes of the pick-and-place machine, this optimization process does not require the control parameters of the easily adjustable variables X and Y to be consistent. Instead, it uses the differentiated motion control parameters of the X and Y axes obtained based on the co-optimization results as part of the optimal set of easily adjustable variable parameters.

[0014] According to the placement accuracy fluctuation optimization method for pick-and-place machines provided by the present invention, based on the multi-machine sample library, a second quantitative relationship model between difficult-to-adjust variables and placement accuracy fluctuation is established using a machine learning algorithm, specifically including: The measured values ​​of difficult-to-adjust variables in the multi-machine sample library are used as input features of the model, and the corresponding mounting accuracy fluctuations are used as outputs of the model. The machine learning algorithm is then used to establish a quantitative relationship between the input features and the output.

[0015] According to the optimization method for placement accuracy fluctuation of a pick-and-place machine provided by the present invention, the machine learning algorithm includes algorithms such as random forest algorithm or gradient boosting decision tree.

[0016] According to the optimization method for placement accuracy fluctuation of a pick-and-place machine provided by the present invention, one or more key difficult-to-adjust variables are identified based on the second quantitative relationship model, specifically including: The contribution of each difficult-to-adjust variable to the fluctuation of mounting accuracy is calculated by analyzing the second quantitative relationship model, and the first preset number of variables with the largest contribution to the fluctuation of mounting accuracy are determined as the key difficult-to-adjust variables.

[0017] According to the pick-and-place machine placement accuracy fluctuation optimization method provided by the present invention, the optimized tolerance range that should be satisfied to achieve the preset placement accuracy fluctuation target is derived in reverse for the key difficult-to-adjust variables, specifically including: Monte Carlo simulation is performed using the second quantitative relationship model. Through iterative analysis, the range of variable values ​​that can make the placement accuracy fluctuation meet the preset target with a high probability is determined, which is used as the optimized tolerance range.

[0018] According to the optimization method for placement accuracy fluctuation of a pick-and-place machine provided by the present invention, the placement accuracy fluctuation is characterized by calculating the standard deviation of the deviation of the placement point in the X and Y directions.

[0019] This invention also provides a system for optimizing placement accuracy fluctuations in a pick-and-place machine, comprising the following modules: The variable classification and filtering module is configured to divide multiple potential variables that affect the fluctuation of mounting accuracy into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment. The adjustable variable optimization unit is configured to obtain placement deviation data under different combinations of adjustable variables through experimental design on a pick-and-place machine; based on the placement deviation data, establish a first quantitative relationship model between the adjustable variables and the placement accuracy fluctuation; and according to the first quantitative relationship model, identify the key adjustable variables that affect the placement accuracy fluctuation, and obtain the optimal set of adjustable variable parameters that minimizes the placement accuracy fluctuation. The difficult-to-adjust variable optimization unit is configured to uniformly apply the optimal easily adjustable variable parameter set on multiple pick-and-place machines of the same model; collect the measured values ​​of the difficult-to-adjust variables and their corresponding placement deviation data for each pick-and-place machine to form a multi-machine sample library; based on the multi-machine sample library, a second quantitative relationship model between the difficult-to-adjust variables and the placement accuracy fluctuation is established using a machine learning algorithm; and based on the second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target is derived in reverse for the key difficult-to-adjust variables.

[0020] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the placement accuracy fluctuation optimization method of any of the above-described methods.

[0021] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the placement accuracy fluctuation optimization method for a pick-and-place machine as described above.

[0022] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the placement accuracy fluctuation optimization method for pick-and-place machines as described above.

[0023] This invention provides a method and system for optimizing placement accuracy fluctuations in a pick-and-place machine. The method categorizes variables affecting accuracy into easily adjustable and difficult-to-adjust variables. Then, through experimental design, a first quantitative relationship model is established between easily adjustable variables and placement accuracy fluctuations, and the optimal software parameter set is determined accordingly. Finally, under the optimal parameters, based on assembly data and measured accuracy from multiple machines, a second quantitative relationship model is established between difficult-to-adjust variables and accuracy fluctuations using machine learning algorithms. This identifies key mechanical tolerances, and simulation technology is used to deduce the optimized tolerance range that meets the accuracy target. Through a systematic data-driven approach, scientific, quantitative, and source-level optimization of placement accuracy fluctuations is achieved. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating the method for optimizing placement accuracy fluctuations in a pick-and-place machine provided by the present invention.

[0026] Figure 2 This is a schematic diagram of the structure of the chip mounter mounting accuracy fluctuation optimization system provided by the present invention.

[0027] Figure 3 This is a technical roadmap for the chip mounter mounting accuracy fluctuation optimization system provided by the present invention.

[0028] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0030] The present invention will now be described in detail with reference to the accompanying drawings. The specific operation methods in the method embodiments can also be applied to the device embodiments or system embodiments. In the description of the present invention, unless otherwise stated, "at least one" includes one or more. "Multiple" refers to two or more. For example, at least one of A, B, and C includes: A existing alone, B existing alone, A and B existing simultaneously, A and C existing simultaneously, B and C existing simultaneously, and A, B, and C existing simultaneously. In the present invention, " / " means "or". For example, A / B can mean A or B. "And / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0031] The present invention will now be described in detail with reference to specific embodiments.

[0032] In some specific embodiments of the present invention, such as Figure 1 As shown, this solution provides a method for optimizing the placement accuracy fluctuation of a pick-and-place machine, including: Step 100: Divide the multiple potential variables that affect the fluctuation of mounting accuracy into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment; Step 200: On a pick-and-place machine, obtain placement deviation data under different combinations of easily adjustable variables through experimental design; based on the placement deviation data, establish a first quantitative relationship model between easily adjustable variables and placement accuracy fluctuation; and according to the first quantitative relationship model, identify the key easily adjustable variables for shadow line placement accuracy fluctuation, and obtain the optimal set of easily adjustable variable parameters that minimizes placement accuracy fluctuation. Step 300: Apply the optimal set of easily adjustable variable parameters uniformly to multiple pick-and-place machines of the same model; collect the measured values ​​of the difficult-to-adjust variables and their corresponding placement deviation data for each pick-and-place machine to form a multi-machine sample library; based on the multi-machine sample library, use a machine learning algorithm to establish a second quantitative relationship model between the difficult-to-adjust variables and the placement accuracy fluctuation; and based on the second quantitative relationship model, identify one or more key difficult-to-adjust variables, and deduce the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target for the key difficult-to-adjust variables.

[0033] It should be noted that existing precision optimization schemes for pick-and-place machines lack a systematic identification, quantification, and collaborative optimization of the numerous factors affecting precision. Motion control parameters largely rely on static calibration based on human experience, failing to adapt to dynamic operating conditions and exhibiting unclear coupling relationships with mechanical assembly errors, leading to a blind and inefficient optimization process. First, the data from mechanical assembly, motion control, and vision inspection systems are fragmented, failing to form a closed-loop feedback loop. In particular, it is impossible to establish a precise quantitative relationship between electromechanical assembly tolerances, control parameters, and final placement precision fluctuations. Second, there is a lack of online compensation mechanisms for dynamic mechanical errors caused by thermal deformation and wear, as well as a lack of precision prediction models that integrate multi-source data, making intelligent self-optimization impossible. Furthermore, existing technologies often use the same control parameters for the X and Y axes with different structures and load characteristics, ignoring the physical differences and motion coupling effects between the two axes, thus limiting dynamic precision under combined motion.

[0034] Therefore, this invention constructs a hybrid optimization system that integrates working mechanisms and experimental data. Through its inherent variable classification and serialization processing mechanism, hybrid modeling method that integrates mechanisms and data, and collaborative optimization strategy oriented towards global accuracy, it achieves systematic identification, quantification, and optimization of multi-source, coupled variables that affect the fluctuation of mounting accuracy.

[0035] In some specific embodiments of the present invention, such as Figure 2 As shown, this solution also provides a pick-and-place machine mounting accuracy fluctuation optimization system to implement the above method, including a variable classification and screening module, an easily adjustable variable optimization unit, and a difficult-to-adjust variable optimization unit.

[0036] Among them, the variable classification and screening module is configured to divide multiple potential variables that affect the fluctuation of mounting accuracy into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment. The adjustable variable optimization unit is configured to obtain placement deviation data under different combinations of adjustable variables through experimental design on a pick-and-place machine; based on the placement deviation data, establish a first quantitative relationship model between the adjustable variables and the placement accuracy fluctuation; and according to the first quantitative relationship model, identify the key adjustable variables that affect the placement accuracy fluctuation, and obtain the optimal set of adjustable variable parameters that minimizes the placement accuracy fluctuation. The difficult-to-adjust variable optimization unit is configured to uniformly apply the optimal easily adjustable variable parameter set on multiple pick-and-place machines of the same model; collect the measured values ​​of the difficult-to-adjust variables and their corresponding placement deviation data for each pick-and-place machine to form a multi-machine sample library; based on the multi-machine sample library, a second quantitative relationship model between the difficult-to-adjust variables and the placement accuracy fluctuation is established using a machine learning algorithm; and based on the second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target is derived in reverse for the key difficult-to-adjust variables.

[0037] Furthermore, such as Figure 3 As shown, Figure 3 This example illustrates the technical roadmap for a pick-and-place machine mounting accuracy fluctuation optimization system. Figure 3 Module 1, Module 2, and Module 3 are respectively Figure 2 The variable classification and filtering module, the easily adjustable variable optimization unit, and the difficult-to-adjust variable optimization unit are discussed below. Figure 3 The implementation process of the method of the present invention is described in detail, and this process fully embodies... Figure 1 The three core steps are given in the text.

[0038] Step 100: Divide the multiple potential variables that affect the fluctuation of mounting accuracy into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment; This step involves variable classification and serialization. First, the multiple potential variables affecting placement accuracy fluctuations are divided into two categories: Adjustable variables: These are variables that can be quickly adjusted by modifying the parameters of the control software, such as the maximum speed of the X / Y / Z axes, motion settling time, and blowing air pressure.

[0039] Difficult-to-adjust variables refer to variables involving mechanical structures that require machine shutdown and adjustment using tools, such as the parallelism of the X / Y / Z axis guides, spline shaft runout, and transmission belt tension. The optimization process of this invention strictly follows the sequential order of "optimizing easily adjustable variables first, then optimizing difficult-to-adjust variables."

[0040] In some possible embodiments of the present invention, after dividing the variables into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment, the method further includes: Construct a network diagram of fluctuation propagation paths that describes the error propagation relationships among various potential influencing variables; Based on the aforementioned fluctuation propagation path network diagram, and combined with expert knowledge, the potential influencing variables are initially screened to determine the list of variables to enter the optimization process.

[0041] In one optional implementation, a network diagram of the fluctuation propagation path describing the error transmission relationship between various variables can be constructed first, such as "spline shaft runout → Z-axis rotational vibration → nozzle end displacement → mounting position deviation". Subsequently, this network diagram is analyzed in conjunction with the knowledge of domain experts to initially screen variables, focusing on core influencing variables and improving the efficiency of subsequent optimization.

[0042] In some possible embodiments of the present invention, the mounting accuracy fluctuation is characterized by calculating the standard deviation of the mounting point in the X and Y directions.

[0043] This invention successfully decouples complex variable relationships by clearly classifying all potential influencing factors into "easily adjustable variables" and "difficult-to-adjust variables" and performing serialization optimization, transforming the optimization process from guesswork based on experience into a scientific and systematic process.

[0044] In one specific embodiment, see still Figure 3 The variable classification and serialization mechanism executed by Module 1, based on engineering mechanisms and adjustment costs, serves as the top-level design for the entire optimization process, ensuring the systematic nature and engineering feasibility of the optimization. Its core components are as follows: Part 1: Analysis of Potential Variables Affecting Mounting Accuracy Fluctuations: This section systematically analyzes all potential influencing variables from multiple dimensions, including "people, machines, materials, methods, and environment," to create a comprehensive and structured variable list, ensuring the completeness of the analysis starting point. The "machine" dimension can be further divided into mechanical systems (e.g., guide rail parallelism, spline shaft runout), motion control systems (e.g., X / Y / Z axis acceleration / deceleration time, X / Y / Z axis maximum speed, settling time), and vision systems (e.g., lighting).

[0045] Part Two: Establishing a Fluctuation Propagation Path Analysis Based on Engineering Mechanisms: The variables obtained are connected based on the working principle of the pick-and-place machine to form a directed network diagram. This network diagram can clearly describe how errors are transmitted and amplified from the source through intermediate variables, ultimately affecting the placement position deviation, thus providing a theoretical basis for subsequent variable selection.

[0046] Part Three: Variable Classification Based on Adjustment Costs and Initial Screening by Experts Based on Mechanisms: Based on the physical characteristics and engineering adjustment costs of variables, clear classification criteria are established, dividing variables into easily adjustable variables (which can be quickly changed through software parameter modification or simple adjustments, such as motion parameters of each axis, settling time, and air pressure) and difficult-to-adjust variables (involving mechanical structures, requiring machine shutdown and tool-assisted adjustment, such as guide rail parallelism and spline shaft runout). Domain experts are then organized to conduct initial screening using engineering knowledge based on the constructed wave propagation path network diagram, focusing on the core variable set to reduce costs and increase efficiency for subsequent experiments.

[0047] Part Four explicitly stipulates that the optimization process must follow a sequential order of "optimizing easily adjustable variables first, and then optimizing difficult-to-adjust variables under their optimal settings." This strategy effectively decouples the complex interactions between variables and avoids distortion of optimization results for easily adjustable variables due to the inherent bias of difficult-to-adjust variables.

[0048] Step 200: On a pick-and-place machine, obtain placement deviation data under different combinations of easily adjustable variables through experimental design; based on the placement deviation data, establish a first quantitative relationship model between easily adjustable variables and placement accuracy fluctuation; and according to the first quantitative relationship model, identify the key easily adjustable variables that affect placement accuracy fluctuation, and obtain the optimal set of easily adjustable variable parameters that minimizes placement accuracy fluctuation. This step is an optimization process using easily adjustable variables.

[0049] In some possible embodiments of the present invention, establishing a first quantitative relationship model between easily adjustable variables and mounting accuracy fluctuations specifically includes: Experiments were conducted using full-factor or partial-factor experimental design methods, and the first quantitative relationship model was established using response surface methodology or multiple linear regression.

[0050] Furthermore, in some possible embodiments of the present invention, the step of identifying key adjustable variables affecting the fluctuation of mounting accuracy based on the first quantitative relationship model, and obtaining the optimal set of adjustable variable parameters that minimizes the fluctuation of mounting accuracy, specifically includes: Analyze the significance of the coefficients of each easily adjustable variable in the first quantitative relationship model; Analysis shows factor plots of the main effects or cross effects of each easily adjustable variable; Based on the factor plot, one or more key adjustable variables that have a significant impact on the fluctuation of mounting accuracy are identified; Based on the first quantitative relationship model, with the optimization objective of minimizing the fluctuation of mounting accuracy, the one or more key adjustable variables are synergistically optimized to solve for the optimal set of adjustable variable parameters.

[0051] In some possible embodiments of the present invention, the step of collaboratively optimizing the one or more key adjustable variables further includes: In view of the differences in mechanical structure, motion inertia and coupling effect between the X and Y axes of the pick-and-place machine, based on the first quantitative relationship model, collaborative optimization is carried out with the comprehensive index of minimizing the placement accuracy fluctuation in the X and Y directions as the optimization objective, so as to obtain an optimal set of easily adjustable variable parameters that includes axially differentiated motion control parameters with different parameter values ​​for the X and Y axes.

[0052] Based on the dynamic characteristic analysis of each axis and actual mounting data, this invention performs collaborative optimization of the motion control parameters of the two axes to obtain differentiated parameter configurations for the two axes. This responds to the essential differences between the two axes in terms of structural rigidity and motion characteristics, and significantly improves the dynamic accuracy and stability of the composite motion trajectory.

[0053] This step uses a dual-head pick-and-place machine as an example. The specific optimization process includes: Experimental design: For the selected easily adjustable variables (such as maximum speed of X / Y / Z axis and settling time), the experimental matrix is ​​constructed using full factorial or partial factorial experimental design (DOE) methods.

[0054] Data Acquisition: Following each parameter combination in the experimental matrix, a standardized placement task was run on the pick-and-place machine. Actual X and Y deviation data for each placement point were collected.

[0055] Characterization of accuracy fluctuation: To quantify the fluctuation of mounting accuracy, the standard deviation of the X-axis deviation was calculated for each experimental group. ) and the standard deviation of the Y-axis deviation ( ), which serves as the core response variable characterizing the fluctuation of mounting accuracy.

[0056] Establish the first quantitative relationship model: Input the experimental data into statistical analysis software, and use methods such as response surface methodology (RSM) or multiple linear regression to fit the fluctuation of mounting accuracy ( The mathematical model for each easily adjustable variable is the first quantitative relationship model.

[0057] Solving for the optimal parameter set: Based on the established first quantitative relationship model, with the goal of minimizing the fluctuation of mounting accuracy, we identify the key adjustable variables that affect the fluctuation of mounting accuracy and obtain the optimal set of adjustable variable parameters.

[0058] As an important preferred embodiment of the present invention, when solving for the optimal parameter set, axial differentiation optimization can be performed on the motion control parameters on the X and Y axes, which are easily adjustable variables. Specifically, considering the differences in mechanical structure and load (the Y axis carries the entire placement head) between the X and Y axes of the pick-and-place machine, their optimal dynamic performance parameters should also be different. Therefore, this step specifically includes: minimizing the comprehensive index of placement accuracy fluctuations in the X and Y directions (e.g., ... To achieve the optimization objective, a collaborative optimization solution is performed to obtain an optimal set of easily adjustable variable parameters that includes axially differentiated motion control parameters with different parameter values ​​for the X and Y axes (e.g., the maximum speed of the X axis is 1.9 m / s and the maximum speed of the Y axis is 1.7 m / s).

[0059] This invention establishes a precise quantitative relationship model between electromechanical parameters and mounting accuracy fluctuations through systematic experimental design and data analysis. This transforms abstract influence relationships into predictable mathematical formulas, providing a direct scientific basis for parameter optimization and tolerance setting.

[0060] In one specific embodiment, see still Figure 3 The mechanism and data fusion-based easily adjustable variable optimization unit in Module 2 is used for rapid response and precise optimization of online adjustable parameters. It mainly includes the following parts: Part 1: Efficient Multivariate Search Based on Experimental Design: Using experimental design methods such as partial factorial and full factorial, a scientific experimental matrix is ​​constructed for multiple easily adjustable variables after screening, so as to efficiently explore the multidimensional parameter space with the fewest number of experiments.

[0061] Part Two: Experimental Data Acquisition and Feature Extraction. Following the experimental matrix, the placement machine simulation program was executed. The deviations of the placement points in the X and Y directions for each experiment were collected, and their standard deviations were calculated as the core response variables. When conditions permitted, vibration signals from each axis and the nozzle were collected to analyze vibration modes under different parameter designs, aiding in parameter optimization.

[0062] Part Three: Establishment of the First Quantitative Relationship Model and Identification of Key Factors: Using tools such as factor plots, key adjustable variables that significantly affect placement accuracy fluctuations were identified from the experimental data. Subsequently, response surface methodology or linear regression models were employed to establish the first quantitative relationship model between the key adjustable variables and placement accuracy fluctuations (either holistically or separately in the X and Y axes). This model clearly reveals the quantitative relationship between parameters and accuracy, such as "Z-axis acceleration is positively correlated with placement accuracy fluctuations."

[0063] Part Four: Optimization and Iterative Verification of Key Adjustable Variables: Based on the quantitative model, and under constraints such as placement efficiency, an optimization algorithm is used to find the optimal parameter combination that minimizes placement accuracy fluctuations. The optimized parameter set is integrated into a control table and uploaded to the pick-and-place machine for iterative verification until satisfactory performance is achieved. Notably, this system supports more refined axial differentiation optimization. Specifically, based on the first quantitative relationship model, a collaborative optimization solution is performed with the goal of minimizing the comprehensive index of placement accuracy fluctuations in the X and Y directions. This yields a set of axially heterogeneous optimal motion control parameters tailored to the differentiated dynamic characteristics of the two axes, addressing control problems caused by structural asymmetry.

[0064] Step 300: Apply the optimal set of easily adjustable variable parameters uniformly to multiple pick-and-place machines of the same model; collect the measured values ​​of the difficult-to-adjust variables and their corresponding placement deviation data for each pick-and-place machine to form a multi-machine sample library; based on the multi-machine sample library, use a machine learning algorithm to establish a second quantitative relationship model between the difficult-to-adjust variables and the placement accuracy fluctuation; and based on the second quantitative relationship model, identify one or more key difficult-to-adjust variables, and deduce the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target for the key difficult-to-adjust variables.

[0065] This step is an optimization process for variables that are difficult to adjust (assembly tolerances).

[0066] In some possible embodiments of the present invention, the machine learning algorithm includes the random forest algorithm or the gradient boosting decision tree algorithm.

[0067] In some possible embodiments of the present invention, the step of establishing a second quantitative relationship model between difficult-to-adjust variables and mounting accuracy fluctuations based on the multi-machine sample library and using machine learning algorithms specifically includes: The measured values ​​of difficult-to-adjust variables in the multi-machine sample library are used as input features of the model, and the corresponding mounting accuracy fluctuations are used as outputs of the model. The machine learning algorithm is then used to establish a quantitative relationship between the input features and the output.

[0068] In some possible embodiments of the present invention, the step of identifying one or more key difficult-to-adjust variables according to the second quantitative relationship model specifically includes: The contribution of each difficult-to-adjust variable to the fluctuation of mounting accuracy is calculated by analyzing the second quantitative relationship model, and the first preset number of variables with the largest contribution to the fluctuation of mounting accuracy are determined as the key difficult-to-adjust variables.

[0069] In some possible embodiments of the present invention, the step of reverse-engineering the optimized tolerance range that should be satisfied to achieve the preset mounting accuracy fluctuation target for the key difficult-to-adjust variable specifically includes: Monte Carlo simulation is performed using the second quantitative relationship model. Through iterative analysis, the range of variable values ​​that can make the placement accuracy fluctuation meet the preset target with a high probability is determined, which is used as the optimized tolerance range.

[0070] For example, to construct a multi-machine sample library, select multiple (e.g., 6) placement machines of the same model. Set all easily adjustable variables of these machines to the optimal parameter set obtained in step two. Then, extract the measured values ​​of difficult-to-adjust variables (such as Z-axis spline runout, guide rail parallelism, etc.) from the assembly quality inspection report of each machine, and run a standard placement task to determine the corresponding overall machine placement accuracy fluctuation. This constitutes a multi-machine sample library.

[0071] Furthermore, a second quantitative relationship model is established. Using the measured values ​​of difficult-to-adjust variables from a multi-machine sample library as input features and the fluctuation of mounting accuracy as the output target, a regression prediction model is trained using machine learning algorithms (such as random forest or gradient boosting decision tree). This is the second quantitative relationship model.

[0072] Based on this, key difficult-to-adjust variables are identified. After the model is trained, by analyzing the ranking of its feature importance, the key difficult-to-adjust variables that contribute the most to the fluctuation of mounting accuracy can be quantitatively identified. For example, the analysis results may show that "spline axis runout" is the factor with the greatest impact on Y-axis accuracy.

[0073] Finally, tolerance inversion optimization is performed. For the identified key difficult-to-adjust variables, Monte Carlo simulation is conducted using a pre-trained second quantitative relationship model. First, a preset placement accuracy fluctuation target is set (e.g., ...). The probability is greater than 99.7%. Then, a large number of random samples are taken within the possible tolerance range of the key, difficult-to-adjust variables, and the probability of the model predicting the accuracy target is input. By iteratively tightening the sampling range (i.e., the tolerance range), an optimized tolerance range that can meet the preset accuracy target with a high probability is finally determined. This range will serve as the new assembly process specification to guide subsequent production.

[0074] In one specific embodiment, see still Figure 3 The multi-machine data mining unit for difficult-to-adjust variables (assembly tolerances) in Module 3 is used to solve the source optimization problem of assembly tolerances, and mainly includes the following parts: Part 1: Construction of a Multi-Machine Sample Library: Select multiple placement machines of the same model. These sample machines should cover the widest possible range of mechanical assembly tolerances, and each machine should have complete records of measured values ​​for difficult-to-adjust variables. On all sample machines, apply the previously obtained optimal set of easily adjustable variable parameters to eliminate interference from differences in control parameters.

[0075] Part Two: Performing Placement Simulation Experiments and Data Acquisition: Run standardized placement test programs on each sample machine, collect placement deviations, and calculate the X and Y direction fluctuations at the whole machine level as responses.

[0076] Part Three: Key Tolerance Identification and Establishment of the Second Quantitative Relationship Model Based on Machine Learning: Subsequently, using the measured values ​​of all difficult-to-adjust variables as input features, machine learning algorithms capable of handling nonlinear relationships, such as random forest and gradient boosting machine, were employed to establish a high-precision quantitative prediction model between the difficult-to-adjust variables and the fluctuation of mounting accuracy as the second quantitative relationship model. By analyzing the contribution of each difficult-to-adjust variable obtained from this model to the fluctuation of mounting accuracy, the top K key difficult-to-adjust variables that contribute the most to the fluctuation of mounting accuracy can be identified quantitatively and with evidence (for example, the results show that "spline axis runout" and "Y-axis guide parallelism" are the most critical factors affecting Y-axis accuracy).

[0077] Part Four: Tolerance Inversion Optimization Design Based on Monte Carlo Simulation: First, based on actual requirements, a desired placement accuracy fluctuation threshold is set. Then, a Monte Carlo simulation is performed using a trained model. A large number of random samples are taken from the identified key, difficult-to-adjust variables within their possible tolerance ranges. Other non-key, difficult-to-adjust variables are fixed according to actual empirical values, and the distribution of placement accuracy fluctuations is observed. Through iterative analysis, the value range of the key, difficult-to-adjust variables that ensures the placement accuracy fluctuations of most simulation results meet the set accuracy target are identified as the optimized tolerances. The obtained optimized tolerance results are integrated into key mechanical assembly tolerance specifications and used as guidance for subsequent pick-and-place machine assembly.

[0078] To further describe the application of this invention in actual processes, the following is in conjunction with... Figure 3 and one A complete embodiment will be provided to illustrate the process of the invention in detail.

[0079] In this embodiment, the goal is to optimize the placement accuracy fluctuation of a dual-head 12-nozzle SMT pick-and-place machine.

[0080] First, a variable classification and serialization mechanism based on engineering mechanisms and adjustment costs specifically includes: Step 1: Assemble a team consisting of data analysts, mechanical engineers, and control engineers.

[0081] Step 2: By analyzing the design drawings and assembly process, and after preliminary screening, identify the variables that may affect the placement accuracy of the pick-and-place machine, including: maximum speed of X / Y / Z axes (easy), settling time (easy), air pressure (easy), guide rail parallelism (difficult), spline shaft runout (difficult), belt tension (difficult), Z-axis no-load tension (difficult), Z-axis no-load thrust (difficult), etc.

[0082] Step 3: Determine the propagation path of the fluctuation, for example: spline shaft runout → Z-axis rotational vibration → horizontal displacement of the nozzle end → placement position deviation.

[0083] Secondly, the role of the easily adjustable variable optimization unit in the fusion of mechanism and data specifically includes: Step 1: For the five selected easily adjustable variables, adopt a two-level full factorial design experimental design scheme. Each parameter combination is run twice. The factor level design of each variable is shown in Table 1. Table 1 is the factor level design table of easily adjustable variables.

[0084] Table 1 Step 2: Select an SMT placement machine and conduct a full factorial experiment according to the factor level design table above. Repeat the experiment twice for each factor combination and collect experimental data. Perform factor plot analysis and linear regression analysis on the experimental data. The results show that, overall, the standard deviation of the placement deviation is positively correlated with the maximum speed along the Z-axis and Y-axis, negatively correlated with the maximum speed along the X-axis, and negatively correlated with the settling time. Furthermore, the influence of Y-axis acceleration is more prevalent than other factors. Based on this, further optimization of the maximum speed and settling time along the X, Y, and Z axes is proposed.

[0085] Step 3: Employ a full-factor experimental design including the center point (e.g., setting the factor level for maximum speed to 20% above and below the empirical maximum speed). With the difficult-to-adjust variables and other easily adjustable variables remaining constant, run different parameter settings to control the placement head and complete the placement simulation, obtaining the placement deviation and calculating the standard deviations of the X and Y direction deviations as response values. Input the experimental data into the program algorithm to fit a quadratic response surface model of the standard deviation with respect to acceleration, and then optimize it. Integrate the optimal parameters at each level to form a preliminary adaptive parameter table.

[0086] Based on experimental data, a first quantitative relationship model was established. This model is a multivariate quadratic response surface model between the fluctuation of mounting accuracy (represented by the standard deviation of each X-axis or Y-axis deviation) and key easily adjustable variables. Its general form can be expressed as: (1), in, This represents the standard deviation of the placement position deviation in the X and Y directions, i.e., the fluctuation of placement accuracy. Indicates the first Key easily adjustable variables, such as wait; For constant terms, The coefficient of the linear term reflects the variable. The main effect, The coefficient of the quadratic term reflects the variable. Nonlinear effects The coefficient of the interaction term reflects the variable and Interaction effect between them This represents the random error term. Experimental data are input into the algorithm to fit the aforementioned quadratic response surface model for each easily adjustable variable, and then optimized. The optimal parameters at each level are integrated to form a preliminary adaptive parameter table.

[0087] Next, the role of the optimization unit for difficult-to-adjust variables (assembly tolerances) based on multi-machine data mining is specifically included in: Step 1: Select 6 dual-head 12-nozzle SMT placement machines as the research sample. From the assembly quality inspection report, identify the difficult-to-adjust variables such as Z-axis spline shaft runout, belt tension, Z-axis no-load thrust, Z-axis no-load tension, and X / Y / Z-axis guide rail parallelism. Set the remaining control variables to the optimal values.

[0088] Step 2: Perform the same mounting task on each sample and repeat it 10 times. Record the deviation of the mounting points in the X and Y directions, and calculate the standard deviation of the deviation in the two directions (mounting accuracy fluctuation).

[0089] Step 3: Establish a model with mounting accuracy fluctuation as the response variable and assembly variable as the independent variable.

[0090] Based on multi-machine sample data, a second quantitative relationship model is constructed. This model uses machine learning algorithms to establish a nonlinear mapping relationship from difficult-to-adjust variables to mounting accuracy fluctuations. With easily adjustable variables fixed, its functional form can be expressed as: (2), in, This represents the standard deviation of the placement position deviation in the X and Y directions, i.e., the fluctuation of placement accuracy. represent Difficult-to-adjust variables include spline shaft runout, Y-axis guide parallelism, and belt tension. It is a complex nonlinear function learned by an algorithm. This is the set of model parameters. For example, when using a gradient boosting tree model, An additive combination of multiple decision trees: (3), in The number of trees; For the first The predicted output of each decision tree; For the first The weights of each decision tree are learned through the training process; model parameters This includes the tree structure, the number of leaf nodes, and weight coefficients.

[0091] After the model training is complete, acid can be added to each difficult-to-adjust variable. Feature Importance Score This allows for the quantitative identification of key influencing factors. The calculation of feature importance scores is based on all decision tree types and uses variables. The reduction in loss function resulting from the partitioning is combined and then normalized: (4), in, Indicates the first All variables used in the decision tree The set of nodes to be split; Indicates at node Splitting reduces the loss function. Importance score. The higher the value, the better the variable. The greater the impact on placement accuracy fluctuations.

[0092] The tolerance inversion optimization process utilizes the trained model described above. Implement the Monte Carlo strategy and set the target accuracy value as follows: Random sampling was performed on the identified key, difficult-to-adjust variables within their possible range. For each sampling, the non-critical, difficult-to-adjust variables and easily-adjustable variables are fixed, and the model is used... Calculate and predict volatility Ultimately, the search for something that can satisfy... The proportion of samples exceeding the range of key variables for the predicted confidence level is used as the optimized tolerance.

[0093] This invention constructs a quantitative regression model of patch variance that includes mechanical assembly parameters and motion control parameters, significantly improving the model's significance. This model provides a scientific basis for achieving precise quantitative control of patch deviation sources. Based on this model, the optimization of key assembly parameters in production can be quantitatively and evidence-based. For example, the analysis results show that the Z-axis linear motor assembly parallelism, spline runout, and tension are all significantly correlated with patch variance, providing a clear direction for improving the assembly process. Simultaneously, the quantitative analysis also reveals the significant impact of easily adjustable experimental variables (such as control parameters like maximum speed and settling time of the X, Y, and Z axes) on patch accuracy, and optimizes these easily adjustable variables based on the quantitative model, making their adjustment more scientific, precise, and effective.

[0094] The mounting accuracy fluctuation optimization method and system provided by this invention constructs a phased and sequential optimization process, divides the variables affecting accuracy into easily adjustable variables (such as software control parameters) and difficult-to-adjust variables (such as mechanical assembly tolerances), and establishes an optimization order of "optimizing easily adjustable variables first, and then optimizing difficult-to-adjust variables".

[0095] In the first stage, for easily adjustable variables, an experimental design method is used to efficiently explore the parameter space and establish a first quantitative relationship model between easily adjustable variables and placement accuracy fluctuations. Based on this model, the optimal set of easily adjustable variable parameters that minimizes placement accuracy fluctuations is scientifically and quantitatively solved. As a preferred approach, this method can also optimize the differentiated parameters of the X / Y axes, that is, obtain an optimal parameter set containing axially differentiated motion control parameters with different parameter values ​​for the X and Y axes, in order to address the asymmetry of the physical characteristics of the two axes.

[0096] In the second stage, addressing the difficult-to-adjust variables, and based on the optimal set of easily adjustable variable parameters obtained in the previous stage, a second quantitative relationship model between the difficult-to-adjust variables and the mounting accuracy fluctuations was established by collecting measured values ​​of mechanical assembly tolerances and overall mounting accuracy fluctuation data from multiple machines of the same model. This model can quantitatively identify the key mechanical tolerances that have the greatest impact on accuracy. Finally, based on this model, Monte Carlo simulation and other methods were used to reverse-engineer the optimized assembly tolerance range that the key mechanical components should follow, starting from the desired final mounting accuracy fluctuation target.

[0097] The pick-and-place machine mounting accuracy fluctuation optimization method and system provided by this invention overcomes the blindness of optimization through a systematic architecture of classification and serialization; by constructing two types of quantitative models, it realizes the transformation from experience-based judgment to data-driven approach; through machine learning and tolerance inversion technology, it realizes data-driven reverse design from the overall machine accuracy target to the assembly tolerance of key components; and by supporting axial differential optimization, it significantly improves the dynamic accuracy under high-speed composite motion.

[0098] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a method for optimizing the placement accuracy fluctuation of a pick-and-place machine. This method includes: classifying multiple potential variables affecting placement accuracy fluctuation into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment; obtaining placement deviation data under different combinations of easily adjustable variables on a pick-and-place machine through experimental design; establishing a first quantitative relationship model between the easily adjustable variables and placement accuracy fluctuation based on the placement deviation data; and identifying key easily adjustable variables affecting placement accuracy fluctuation according to the first quantitative relationship model. The optimal set of easily adjustable variable parameters that minimizes placement accuracy fluctuations is selected. This optimal set of easily adjustable variable parameters is uniformly applied to multiple placement machines of the same model. The measured values ​​of difficult-to-adjust variables and their corresponding placement deviation data are collected for each placement machine to form a multi-machine sample library. Based on the multi-machine sample library, a second quantitative relationship model between difficult-to-adjust variables and placement accuracy fluctuations is established using a machine learning algorithm. Based on the second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target is derived in reverse for the key difficult-to-adjust variables.

[0099] Furthermore, when the logical instructions in the aforementioned memory 430 are implemented as functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0100] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the placement accuracy fluctuation optimization method for pick-and-place machines provided by the above methods. This method includes: classifying multiple potential influencing variables affecting placement accuracy fluctuation into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment; obtaining placement deviation data under different combinations of easily adjustable variables on a pick-and-place machine through experimental design; and establishing a first quantitative relationship model between easily adjustable variables and placement accuracy fluctuation based on the placement deviation data; and Based on the first quantitative relationship model, key adjustable variables affecting placement accuracy fluctuations are identified, and the optimal adjustable variable parameter set that minimizes placement accuracy fluctuations is obtained. This optimal adjustable variable parameter set is uniformly applied to multiple placement machines of the same model. Measured values ​​of difficult-to-adjust variables and their corresponding placement deviation data are collected from each placement machine to form a multi-machine sample library. Based on this multi-machine sample library, a second quantitative relationship model between difficult-to-adjust variables and placement accuracy fluctuations is established using a machine learning algorithm. Furthermore, based on the second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be satisfied to achieve the preset placement accuracy fluctuation target is derived in reverse for these key difficult-to-adjust variables.

[0101] Furthermore, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the optimization method for placement accuracy fluctuation of a pick-and-place machine provided by the methods described above. This method includes: classifying multiple potential variables affecting placement accuracy fluctuation into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment; obtaining placement deviation data under different combinations of easily adjustable variables on a pick-and-place machine through experimental design; establishing a first quantitative relationship model between the easily adjustable variables and placement accuracy fluctuation based on the placement deviation data; and obtaining... The optimal set of easily adjustable variable parameters that minimizes placement accuracy fluctuations is identified, and key easily adjustable variables affecting placement accuracy fluctuations are identified. This optimal set of easily adjustable variable parameters is uniformly applied to multiple placement machines of the same model. Measured values ​​of difficult-to-adjust variables and their corresponding placement deviation data are collected from each placement machine to form a multi-machine sample library. Based on this multi-machine sample library, a second quantitative relationship model between difficult-to-adjust variables and placement accuracy fluctuations is established using machine learning algorithms. Based on this second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be satisfied to achieve the preset placement accuracy fluctuation target is derived in reverse for these key difficult-to-adjust variables.

[0102] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0103] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing placement accuracy fluctuations in a pick-and-place machine, characterized in that, include: The multiple potential variables affecting the fluctuation of mounting accuracy are divided into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment. On a pick-and-place machine, placement deviation data under different combinations of easily adjustable variables are obtained through experimental design; based on the placement deviation data, a first quantitative relationship model between easily adjustable variables and placement accuracy fluctuation is established; and according to the first quantitative relationship model, key easily adjustable variables affecting placement accuracy fluctuation are identified, and the optimal set of easily adjustable variable parameters that minimizes placement accuracy fluctuation is obtained. On multiple pick-and-place machines of the same model, the optimal set of easily adjustable variable parameters is uniformly applied; the measured values ​​of the difficult-to-adjust variables and their corresponding placement deviation data are collected for each pick-and-place machine to form a multi-machine sample library; based on the multi-machine sample library, a second quantitative relationship model between the difficult-to-adjust variables and the placement accuracy fluctuation is established using a machine learning algorithm; and based on the second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target is derived in reverse for the key difficult-to-adjust variables.

2. The method for optimizing placement accuracy fluctuation of a pick-and-place machine according to claim 1, characterized in that, After classifying variables into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment, the method further includes: Construct a network diagram of fluctuation propagation paths that describes the error propagation relationships among various potential influencing variables; Based on the aforementioned fluctuation propagation path network diagram, and combined with expert knowledge, the potential influencing variables are initially screened to determine the list of variables to enter the optimization process.

3. The method for optimizing placement accuracy fluctuation in a pick-and-place machine according to claim 1, characterized in that, The establishment of the first quantitative relationship model between easily adjustable variables and mounting accuracy fluctuations specifically includes: Experiments were conducted using full-factor or partial-factor experimental design methods, and the first quantitative relationship model was established using response surface methodology or multiple linear regression.

4. The method for optimizing placement accuracy fluctuations in a pick-and-place machine according to claim 1 or 3, characterized in that, The step of identifying key adjustable variables affecting placement accuracy fluctuations based on the first quantitative relationship model, and obtaining the optimal set of adjustable variable parameters that minimizes placement accuracy fluctuations, specifically includes: Analyze the significance of the coefficients of each easily adjustable variable in the first quantitative relationship model; Analysis shows factor plots of the main effects or cross effects of each easily adjustable variable; Based on the factor plot, one or more key adjustable variables that have a significant impact on the fluctuation of mounting accuracy are identified; Based on the first quantitative relationship model, with the optimization objective of minimizing the fluctuation of mounting accuracy, the one or more key adjustable variables are synergistically optimized to solve for the optimal set of adjustable variable parameters.

5. The method for optimizing placement accuracy fluctuation in a pick-and-place machine according to claim 4, characterized in that, The collaborative optimization of the one or more key adjustable variables further includes: In view of the differences in mechanical structure, motion inertia and coupling effect between the X and Y axes of the pick-and-place machine, based on the first quantitative relationship model, collaborative optimization is carried out with the comprehensive index of minimizing the placement accuracy fluctuation in the X and Y directions as the optimization objective, so as to obtain an optimal set of easily adjustable variable parameters that includes axially differentiated motion control parameters with different parameter values ​​for the X and Y axes.

6. The method for optimizing placement accuracy fluctuation in a pick-and-place machine according to claim 1, characterized in that, The second quantitative relationship model between difficult-to-adjust variables and mounting accuracy fluctuations, based on the multi-machine sample library and using machine learning algorithms, specifically includes: The measured values ​​of difficult-to-adjust variables in the multi-machine sample library are used as input features of the model, and the corresponding mounting accuracy fluctuations are used as outputs of the model. The machine learning algorithm is then used to establish a quantitative relationship between the input features and the output.

7. The method for optimizing placement accuracy fluctuation in a pick-and-place machine according to claim 6, characterized in that, Based on the second quantitative relationship model, one or more key difficult-to-adjust variables were identified, specifically including: Based on the second quantitative relationship model, calculate the contribution of each difficult-to-adjust variable to the fluctuation of mounting accuracy; The variable with the largest pre-set number of factors contributing to the fluctuation of mounting accuracy is identified as the key difficult-to-adjust variable.

8. The method for optimizing placement accuracy fluctuation of a pick-and-place machine according to claim 1, characterized in that, The reverse derivation of the optimized tolerance range that should be met to achieve the preset mounting accuracy fluctuation target for the key, difficult-to-adjust variables specifically includes: Monte Carlo simulation is performed using the second quantitative relationship model. Through iterative analysis, the range of variable values ​​that can make the placement accuracy fluctuation meet the preset target with a high probability is determined, which is used as the optimized tolerance range.

9. The method for optimizing placement accuracy fluctuation of a pick-and-place machine according to claim 1 or 6, characterized in that, The machine learning algorithms include random forest algorithm or gradient boosting decision tree algorithm.

10. The method for optimizing placement accuracy fluctuation of a pick-and-place machine according to claim 1, characterized in that, The fluctuation in mounting accuracy is characterized by calculating the standard deviation of the mounting points in the X and Y directions.

11. A system for optimizing placement accuracy fluctuations in a pick-and-place machine, characterized in that, include: The variable classification and filtering unit is configured to divide multiple potential variables that affect the fluctuation of mounting accuracy into easily adjustable variables that can be parameterized and difficult-to-adjust variables that require mechanical adjustment. The adjustable variable optimization unit is configured to obtain placement deviation data under different combinations of adjustable variables through experimental design on a pick-and-place machine; based on the placement deviation data, establish a first quantitative relationship model between the adjustable variables and the placement accuracy fluctuation; and according to the first quantitative relationship model, identify the key adjustable variables that affect the placement accuracy fluctuation, and obtain the optimal set of adjustable variable parameters that minimizes the placement accuracy fluctuation. The difficult-to-adjust variable optimization unit is configured to uniformly apply the optimal easily adjustable variable parameter set on multiple pick-and-place machines of the same model; collect the measured values ​​of the difficult-to-adjust variables and their corresponding placement deviation data for each pick-and-place machine to form a multi-machine sample library; based on the multi-machine sample library, a second quantitative relationship model between the difficult-to-adjust variables and the placement accuracy fluctuation is established using a machine learning algorithm; according to the second quantitative relationship model, one or more key difficult-to-adjust variables are identified, and the optimized tolerance range that should be met to achieve the preset placement accuracy fluctuation target is derived in reverse for the key difficult-to-adjust variables.

Images