A method and system for dispensing control for discontinuous paths

Abstract

The application discloses a kind of point glue control method and system for discontinuous path, method includes: obtaining three-dimensional point glue track containing at least one path breakpoint, and the track segment before and after path breakpoint is identified, and is recorded as front track segment and rear track segment;Based on the termination geometric feature of front track segment, the starting geometric feature of rear track segment and the space span of path breakpoint, the fracture and lap behavior of glue at path breakpoint is predicted, and special compensation model for controlling the form of glue line at both ends of path breakpoint is established;In the movement of point glue head along front track segment to the front of path breakpoint, according to the special compensation model, the first preset action sequence is executed.The application realizes the intelligent closed-loop control of discontinuous path point glue, effectively solves the form out-of-control problem of glue when crossing physical gap, significantly improves the point glue quality and process adaptive ability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of motor manufacturing, in particular to a point control method and system for discontinuous paths. BACKGROUND

[0002] In the modern motor manufacturing field, precise glue coating of stator assembly is one of the key processes to ensure product structural stability and electrical insulation performance.

[0003] With the popularity of high-performance motors such as brushless DC motors and shielded motors, their stator structures are becoming increasingly complex, often containing core laminations, tooth slots, windings, and insulation skeletons, etc. multiple special-shaped structures, resulting in frequent physical barriers to the glue dispensing path, forming a large number of discontinuous path breakpoints.

[0004] Traditional continuous path glue dispensing methods have inherent limitations when facing such breakpoints. The glue often experiences uncontrolled stringing, dripping when crossing gaps, or difficulty in initial spreading after the breakpoint. This not only affects the appearance, but also can cause serious quality problems, such as contamination of precise electrical areas by excess glue, or increased vibration, stress concentration, and reduced structural strength due to poor glue overlap during high-speed motor operation, directly affecting the reliability, efficiency, and lifespan of the motor.

[0005] Therefore, there is an urgent need in the art for an automated glue dispensing technology that can intelligently adapt to such complex discontinuous paths, ensuring precise, stable, and reliable glue line overlap at each breakpoint. SUMMARY

[0006] To solve the above problems, an embodiment of the present application provides a point control method for discontinuous paths, the method comprising:

[0007] Obtaining a three-dimensional glue dispensing trajectory containing at least one path breakpoint, and identifying the trajectory segments before and after the path breakpoint, denoted as the front trajectory segment and the rear trajectory segment;

[0008] Based on the termination geometric features of the front trajectory segment, the starting geometric features of the rear trajectory segment, and the spatial span of the path breakpoint, predicting the breaking and overlapping behavior of the glue at the path breakpoint, and establishing a special compensation model for controlling the form of the glue line at both ends of the path breakpoint;

[0009] Before the glue dispensing head moves along the front trajectory segment to the path breakpoint, a first preset action sequence is executed according to the special compensation model to form a front glue line with a target ending shape; when the glue dispensing head crosses the path breakpoint and is about to start the rear trajectory segment, a second preset action sequence is executed according to the special compensation model to form a starting shape at the beginning of the rear trajectory segment that matches the ending shape of the front glue line;

[0010] Based on the physical data synchronously collected during the execution of the action sequence and the actual three-dimensional morphology of the glue line subsequently obtained through the visual detection system, a performance error is generated by comparing with the ideal morphology predicted by the special compensation model. The performance error is used to fine-tune the parameters of the special compensation model through an incremental learning algorithm, and the process data of this round and the optimized model parameters are archived to the process knowledge base.

[0011] Further, the special compensation model outputs a decision by comprehensively analyzing the dynamic balance relationship between the spatial span of the path breakpoint and the cohesion and extensional viscosity of the glue, the decision including an in-air wire bonding mode and a substrate reserved bonding mode; when the decision is the in-air wire bonding mode, the glue wire is controlled to cross the breakpoint in a natural parabolic trajectory affected by air resistance and gravity, and the end point of the glue wire overlaps with the starting area of the subsequent trajectory segment within a preset range; when the decision is the substrate reserved bonding mode, a glue anchor point with a predetermined volume and shape is formed at the end of the previous trajectory segment before the point glue head reaches the path breakpoint, and the anchor point serves as a physical basis for bonding at the start of the subsequent trajectory segment.

[0012] Further, the operation mechanism of the special compensation model includes taking the spatial vector of the path breakpoint, the rheological parameters of the glue, and the dynamic crossing speed of the point glue head as inputs, and synchronously calculating the amplitude and speed of the pullback action, and the frequency characteristics and energy level of the high-frequency micro-vibration through a built-in dynamic function representing the viscoelastic properties of the glue.

[0013] Further, the spatial motion path followed by the point glue head when crossing the path breakpoint is a three-dimensional spline curve dynamically generated according to the tension optimization target output by the special compensation model, which ensures the stability of the tension of the glue wire during the crossing process and the accuracy of the landing point.

[0014] Further, the performance error is decoupled into a fracture morphology error corresponding to the end shape of the previous glue line and a spreading morphology error corresponding to the start shape of the subsequent glue line; the incremental learning algorithm selectively reinforces the learning of the sub-modules in the special compensation model that control the pullback action and the high-frequency micro-vibration, respectively, according to the contribution degrees of the fracture morphology error and the spreading morphology error with unequal weights.

[0015] Further, the special compensation model also takes the wetting characteristics of the substrate surface as one of the input parameters for dynamic decision-making; when the model identifies that the substrate is a low surface energy material, the energy level of the high-frequency micro-vibration will be dynamically increased to overcome the energy barrier of the glue when spreading at the start of the subsequent trajectory segment.

[0016] Further, the incremental learning algorithm will start a meta-optimization sub-process after completing the parameter fine-tuning of the current dedicated compensation model; the sub-process first encodes the geometric features, colloid properties and substrate wetting characteristics of the current path discontinuity into a unified feature vector; then, based on this feature vector, it retrieves K most similar historical process cases from the process knowledge base, where K is a pre-set positive integer; further, by analyzing the correlation between the initial model parameters and the final optimized performance in the cases, it optimizes a globally shared model initialization strategy library; the optimized strategy library is used to directly generate a set of initial model parameters better than the default values for new types of discontinuities encountered subsequently.

[0017] Further, the frequency characteristics of the high-frequency micro-amplitude vibration are set within a specific range, which aims to stimulate the thixotropic effect of the colloid, so as to achieve faster spreading and shaping at the beginning of the post-trajectory segment.

[0018] A point dispensing control method for discontinuous paths, the method further comprising:

[0019] Establishing and maintaining a local cache of discontinuity compensation strategies.

[0020] A point dispensing control system for discontinuous paths, the system comprising:

[0021] A trajectory analysis module that obtains a three-dimensional point dispensing trajectory containing at least one path discontinuity, and identifies the trajectory segments before and after the path discontinuity, denoted as the pre-trajectory segment and the post-trajectory segment;

[0022] A behavior modeling module that predicts the breaking and bridging behavior of the colloid at the path discontinuity based on the termination geometric features of the pre-trajectory segment, the starting geometric features of the post-trajectory segment, and the spatial span of the path discontinuity, and establishes a dedicated compensation model for controlling the morphology of the glue line at both ends of the path discontinuity;

[0023] A dynamic execution module that, before the point dispensing head moves along the pre-trajectory segment to the path discontinuity, executes a first pre-set action sequence according to the dedicated compensation model to form a pre-glue line with a target end shape; when the point dispensing head crosses the path discontinuity and is about to start the post-trajectory segment, executes a second pre-set action sequence according to the dedicated compensation model to form a starting shape at the beginning of the post-trajectory segment that is adapted to the end shape of the pre-glue line;

[0024] An online evolution module compares the actual three-dimensional morphology of the glue line obtained through visual detection system after the action sequence is performed with the ideal morphology predicted by the special compensation model based on the physical data synchronously collected when the action sequence is performed, generates performance error, and uses the performance error to fine-tune the parameters of the special compensation model through an incremental learning algorithm, and archives the process data of this round and the optimized model parameters to the process knowledge base.

[0025] The technical effect and advantages of the point gluing control method for discontinuous paths provided by the application are as follows:

[0026] The application realizes intelligent closed-loop control of point gluing of discontinuous paths, effectively solves the problem of form out of control of the glue when crossing the physical gap, and significantly improves the point gluing quality and process self-adaptive ability. The traditional method can only passively deal with the consequences such as stringing and glue dripping when encountering path discontinuity, while the application can predict the breaking and lapping behavior of the glue in advance through the establishment of a special compensation model, and plan accurate compensation actions, thereby fundamentally avoiding the generation of quality defects; by dynamically generating a path crossing action and cooperatively controlling the point gluing action, and introducing online learning and a knowledge base, the system can generate an optimal strategy for different characteristic discontinuities, ensure that the glue line lapping at each discontinuity is stable and reliable, thereby ensuring the structural strength and long-term reliability of the final product (such as a motor stator); not only the geometric path is considered, but also complex factors such as the rheological properties of the glue and the surface energy of the substrate are taken into account, so that the point gluing process can intelligently adapt to changes in materials and environment, thereby reducing the debugging cost and dependence on the experience of the operator when switching objects. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 A point gluing control method flowchart for discontinuous paths in embodiment one;

[0028] Figure 2 A point gluing control method flowchart for discontinuous paths in embodiment two;

[0029] Figure 3 A connection diagram of a point gluing control system for discontinuous paths in embodiment three. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0031] Embodiment one:

[0032] Please refer to Figure 1 Embodiments of the present application provide a point dispensing control method for discontinuous paths, characterized by comprising the following steps:

[0033] S1: Obtain a three-dimensional point dispensing trajectory containing at least one path discontinuity point, and identify the trajectory segments before and after the path discontinuity point, denoted as the front trajectory segment and the rear trajectory segment;

[0034] S2: Based on the termination geometric characteristics of the front trajectory segment, the starting geometric characteristics of the rear trajectory segment, and the spatial span of the path discontinuity point, predict the breaking and joining behavior of the adhesive at the path discontinuity point, and establish a special compensation model for controlling the shape of the adhesive line at both ends of the path discontinuity point;

[0035] S3: Before the dispensing head moves along the front trajectory segment to the path discontinuity point, execute a first preset action sequence according to the special compensation model to form a front adhesive line with a target end shape; when the dispensing head crosses the path discontinuity point and is about to start the rear trajectory segment, execute a second preset action sequence according to the special compensation model to form a starting shape at the start of the rear trajectory segment that matches the end shape of the front adhesive line; wherein the first preset action sequence at least includes pre-reducing the adhesive output before moving to the discontinuity point, and executing a quick pull-back or lifting action at the precise discontinuity point coordinate to actively break the adhesive filament and control its breaking point shape; the second preset action sequence at least includes pre-establishing the adhesive pressure before reaching the start point of the rear trajectory segment, and simultaneously triggering a high-frequency micro-vibration at the moment of contacting the start point;

[0036] S4: Based on the physical data (such as pressure, position, vibration feedback) synchronously collected when executing the action sequence in S3 and the actual three-dimensional topography of the adhesive line subsequently obtained by the visual detection system, compare with the ideal topography predicted by the model in S2 to generate a performance error, use the performance error to fine-tune the parameters of the special compensation model through incremental learning algorithm, and archive this round of process data and the optimized model parameters to the process knowledge base, completing a self-performance evolution for this specific discontinuity point working condition.

[0037] The special compensation model in S2 is not a static lookup table, but an intelligent decision-making subsystem that can comprehensively analyze path geometry, material properties, and dynamic processes.

[0038] The special compensation model outputs the final joining mode decision by comprehensively analyzing the dynamic balance relationship between the spatial span of the path discontinuity point and the cohesive force and tensile viscosity of the adhesive. In essence, this dynamic balance relationship refers to the competition between the internal resistance to breaking of the adhesive and the external force causing its breaking when it is stretched, which directly determines the yield rate of the motor stator point dispensing.

[0039] The spatial span of an inter-path break point refers to the straight-line distance and spatial vector between the end point of the former trajectory segment and the start point of the latter trajectory segment in three-dimensional space, which is a key geometric constraint.

[0040] The cohesion of a gel is the force of attraction between the molecules within a liquid that causes the surface of the liquid to contract and resist being pulled apart.

[0041] The extensional viscosity of a gel describes the ability of the gel to resist flow deformation when subjected to extensional stress, and a gel with high extensional viscosity is more likely to form a long and stable filament.

[0042] The dedicated compensation model simulates a scenario in which the gel can be stretched into an unbroken filament and successfully bridged to the other end when the dispensing head attempts to span a given spatial span, and if so, it tends to adopt the mid-air filament bridging mode; if not, it needs to adopt the base reservation bridging mode.

[0043] The mid-air filament bridging mode includes:

[0044] When the dedicated compensation model determines that the cohesion and extensional viscosity of the gel are sufficient to support it across the current inter-break point spatial span, it will decide on the mid-air filament bridging mode.

[0045] In this mode, the system precisely controls the deceleration of the dispensing head before the inter-break point, the nonlinear reduction of the glue amount, and the final lifting action, and the purpose of this series of actions is to actively guide the gel to break and control the breaking point to form a sharp ending. At the same time, the dispensing head will span the inter-break point at a specific speed trajectory, so that the pulled-out glue filament forms a natural parabolic trajectory under the combined influence of air resistance and gravity; this trajectory is not formed randomly, but is pre-calculated through fluid dynamics simulation to ensure that the end point of the glue filament can form a preset range of overlap with the starting area of the latter trajectory segment. The "preset range" here is a relative concept, for example, it can be set to 50% to 150% of the glue line width to ensure sufficient contact area for reliable electrical or mechanical connection.

[0046] Example: When applying sealant to the slot of a brushless DC motor stator, the glue path needs to span from the side of one stator tooth to the adjacent tooth, assuming the spatial span here is 1.5mm, the dedicated compensation model determines that it meets the "mid-air filament" condition based on this span and the parameters of the epoxy sealant used; the system then executes the predetermined action: the dispensing head nonlinearly reduces the glue amount before reaching the inter-break point, and performs a rapid pullback action at the precise coordinates, which causes the gel to be stretched into a fine and uniform filament and successfully bridged to the starting coating position of the adjacent stator tooth in a natural parabolic trajectory, achieving efficient and clean spanning.

[0047] The base-reserve joint mode includes:

[0048] When the special compensation model determines that the current spatial span or process condition is not suitable for air drawing, for example, it needs to span the top of the motor winding coil or the short-circuit ring fixing column of the shield motor, the space span at these positions may reach 3 mm or more, the model will decide the base-reserve joint mode; the core strategy of this mode is to give up the high-risk air drawing and instead ensure the reliability of the connection by increasing the volume of the base adhesive; the system will not try to make the adhesive complete the span in the air, but will start to prepare the "ground" connection when the front trajectory segment is about to end but has not yet reached the discontinuity; specifically, the model will instruct the glue head to form a local adhesive anchor point with a predetermined volume and shape at the end of the front trajectory segment through a short pause, a small amount of glue increase, etc.; the function of this adhesive anchor point is similar to a pier or bridge pier, which serves as a reliable physical basis, waits for the glue head to span the discontinuity (at this time, there is no need to draw, it may be directly moved or lifted), and then starts to glue at the beginning of the rear trajectory segment, and makes the new glue line fully contact and fuse with the previously reserved adhesive anchor point, thereby physically realizing the continuity of the glue line.

[0049] Example: On the stator core of a shield motor, there is a glue path that needs to bypass the short-circuit ring mounting seat. Before the glue path is broken by the mounting seat, the model will instruct the glue head to form an adhesive anchor point with a predetermined volume and shape at the end of the path, and this anchor point serves as a solid physical basis for the glue head to safely cross the mounting seat and start gluing again on the other side, and the new glue line fully fuses with the reserved adhesive anchor point.

[0050] The operation mechanism of the special compensation model includes: taking the spatial vector of the discontinuity, the rheological parameters of the adhesive, and the dynamic span speed of the glue head as inputs, and through a built-in dynamic function representing the viscoelastic properties of the adhesive, the amplitude and speed of the back-pulling action, and the frequency characteristics and energy level of the high-frequency micro-vibration are simultaneously calculated.

[0051] This operation mechanism ensures that each micro-motion in the selected joint mode perfectly matches the real-time state of the adhesive and the spatial geometry; the operation of the special compensation model is not isolated, but closely depends on a multi-dimensional input set, which collectively defines the current glue dispensing conditions.

[0052] The spatial vector of the discontinuity is not just a distance scalar, but a three-dimensional vector that includes direction; for example, in motor stator dispensing, an upward spatial vector (such as spanning the winding) and a horizontal spatial vector (such as spanning the core tooth slot) have completely different control strategies even if the distance is the same, because the direction of the effect of gravity changes.

[0053] The rheological parameters of a gel are the key physical properties that describe the deformation and flow behavior of a gel, mainly including viscosity (the ability to resist flow) and elasticity (the ability to recover after deformation), which together constitute the viscoelastic properties of a gel, and determine the response of a gel when it is stretched (back-pulling action) and forced to spread (high-frequency micro-vibration).

[0054] The dynamic crossing speed of the dispensing head refers to the movement speed of the dispensing head when passing through the path discontinuity point area, which directly affects the stretching rate applied to the gel, and is one of the core dynamic variables that determine the fracture morphology and length of the gel filament.

[0055] The operation core of the special compensation model is an internal dynamic function that represents the viscoelastic properties of the gel. This function can be understood as a "digital brain" trained with a large amount of data and physical laws, which can simulate the dynamic behavior of the gel under the above input conditions. Its core task is to perform synchronous calculation, i.e., to calculate the control instructions of multiple key actuators at one time.

[0056] Calculate the Profile of the back-pulling action; the model output is not a single back-pulling distance, but a complete profile of amplitude and speed. Profile here refers to a sequence of instructions that changes over time, which precisely specifies the acceleration, constant speed segment, and deceleration of the back-pulling action, thereby controlling the gel fracture process to be "clean and sharp", rather than "dragging and lingering". A steep, high-speed back-pulling action profile is suitable for low-viscosity, easily-flowing gels to quickly cut off the gel filament; while for high-elasticity gels, a profile containing a short pause may be needed to fully utilize their elastic retraction.

[0057] Calculate the parameters of high-frequency micro-vibration; at the same time, the model also calculates the vibration parameters for the second preset action sequence, including frequency characteristics (such as the main frequency value and its harmonic distribution) and energy level (such as amplitude or driving power); the core function of high-frequency micro-vibration is to use vibration energy to instantaneously reduce the apparent viscosity of the gel, so that it can quickly spread out when it contacts the substrate, forming a perfect initial shape and seamlessly merging with the end of the pre-gel line or the reserved gel anchor point.

[0058] Example: consider the scenario of dispensing across the stator of a brushless DC motor between two adjacent teeth.

[0059] Assuming the spatial vector of the discontinuity point is a horizontal vector with a length of 2mm, and the used glue is a silicone glue with certain elasticity, after receiving these inputs, the dynamic function inside the special compensation model starts to calculate, and it can calculate a profile of the pullback action: in 5ms, first start with an acceleration of 50mm / s², and then pull back 0.4mm at a speed of 100mm / s in a very short time; this specific combination can ensure that when the silicone glue is pulled off, a thin and strong wire is formed; at the same time, the model will calculate the vibration parameters with a main frequency of 800Hz and an amplitude of 10 microns for high-frequency micro-vibration, which can effectively overcome the elasticity of the glue and make it form a wide and uniform glue line starting point on the stator tooth surface immediately.

[0060] Through this highly integrated operation mechanism, the present application realizes accurate control of complex fluids in dynamic processes.

[0061] In S3, the spatial motion path followed by the dispensing head when crossing the path discontinuity point is a three-dimensional spline curve dynamically generated according to the tension optimization target output by the special compensation model, which ensures the stability of the tension and the accuracy of the landing point of the glue wire during the crossing process.

[0062] In the air wire drawing lap joint mode, the whole process from the dispensing head leaving the front trajectory segment to reaching the starting point of the rear trajectory segment is the key stage to determine the success or failure of the lap joint; the special compensation model outputs a key physical target, i.e. the tension optimization target, while deciding and outputting the action sequence parameters.

[0063] The tension optimization target refers to maintaining a relatively stable and optimal internal tension during the process of the glue wire being stretched across; if the tension is too small, the glue wire may touch the underlying structure (such as the motor winding) due to excessive sagging under gravity, causing pollution; if the tension is too large, the glue wire may be pulled off before reaching the landing point. The target aims to find a balance point so that the glue wire is neither broken nor slack.

[0064] In order to achieve the above target, the model will calculate and generate a smooth spatial motion path in real time for the dispensing head to execute; this path is usually composed of a three-dimensional spline curve, which is a very smooth curve defined by control points, and its advantage is that it can realize smooth acceleration and speed transition, avoiding additional impact force on the glue wire due to the sudden start and stop of the dispensing head.

[0065] This dynamically generated three-dimensional spline curve is not an arbitrary trajectory, and it has two core physical guarantees:

[0066] By carefully designing the curvature variation of the curve, the speed of the dispensing head during the crossing process can be matched with the rheological properties of the glue, so as to ensure that the tension inside the glue filament is maintained within a preset optimal range; stable tension is the premise of forming uniform glue filament diameter and accurately controlling its parabolic shape.

[0067] The end point of the curve in space is strictly constrained within the starting area of the rear trajectory segment; the model calculates the path that the dispensing head should follow by comprehensively considering the dynamic crossing speed of the dispensing head, gravitational acceleration and air resistance, so that when the dispensing head reaches the end point, the glue filament pulled behind it can accurately fall into the expected lap position under the action of gravity and inertia.

[0068] For example: in a scenario where the motor stator winding needs to be crossed, the spatial span of the path breakpoint is 3 mm, and there is a 2 mm drop; if the dispensing head simply moves along a straight line, due to the gravitational acceleration, the glue filament will be rapidly stretched, the tension will suddenly increase, and it will be easily broken. At this time, the special compensation model will calculate the strategy of "first quickly lifting to accumulate the glue filament, and then slowly descending to release the glue filament" according to the extension viscosity of the glue, and accordingly, the model will generate a specific three-dimensional spline curve: the dispensing head first moves upward at a relatively steep angle for about 1 mm, which helps to buffer the tension and avoid instantaneous breakage; then it moves downward along a smooth curve to the starting point of the rear trajectory segment; this path ensures that the tension of the glue filament changes smoothly during the entire crossing process, and its end accurately falls into the preset lap area with an error of less than ±0.05 mm, thereby achieving high-reliability airborne lapping.

[0069] Through this intelligent optimization of the crossing path itself, the application ensures that even in complex three-dimensional space, the lapping behavior of the glue filament is highly controllable and predictable, thereby significantly improving the yield and connection reliability of dispensing glue for precision components such as brushless DC motors.

[0070] In S4, the performance error is decoupled into a breakage pattern error corresponding to the front glue line end shape and a spreading pattern error corresponding to the rear glue line start shape; the incremental learning algorithm selectively reinforces learning of sub-modules in the special compensation model that control the pullback action and the high-frequency micro-vibration, respectively, with unequal weights according to the contribution of the breakage pattern error and the spreading pattern error.

[0071] In the S4 step, after comparing the actual three-dimensional morphology of the glue line collected by the system with the ideal morphology predicted by the model, the performance error generated is not a single evaluation index. Based on the deep understanding of the glue line morphology, the system intelligently decouples the total error into two independent error components that are respectively linked to the core actions:

[0072] Breakage morphology error: This error component specifically evaluates the deviation of the end shape of the front glue line from the target, for example, an ideal end might be a sharp break, while the actual formation might be a trailing "tail"; this error mainly reflects the control accuracy of the first preset action sequence (especially the pullback action).

[0073] Spread morphology error: This error component specifically evaluates the deviation of the initial shape of the rear glue line from the target, for example, an ideal initial should be a smooth "meniscus" shape, while the actual might be an irregular ball; this error mainly reflects the effectiveness of the second preset action sequence (especially the high-frequency micro-vibration).

[0074] Through this decoupling, the system can diagnose the main contradiction of the current dispensing problem, that is, whether the glue filament is not broken enough, or the initial spread is not sufficient.

[0075] After obtaining the above two error components, the incremental learning algorithm implements a selective reinforcement learning strategy, the core logic of which is: according to the contribution of the breakage morphology error and the spread morphology error, the sub-modules in the special compensation model that control the pullback action and the high-frequency micro-vibration are optimized with unequal weights, which means that if the breakage morphology error is dominant (i.e. the end shape is poor, but the initial shape is acceptable), the algorithm will determine that the problem lies in the filament breakage link, so it will focus on optimizing the part of the model that calculates the amplitude and speed profile of the pullback action, for example, it may significantly increase the initial acceleration weight of the pullback action to pursue a more crisp break; conversely, if the spread morphology error is dominant (i.e. the end shape is good, but the initial shape is a mess), the algorithm will focus on optimizing the frequency characteristics and energy level of the high-frequency micro-vibration control module in the model, such as trying to increase the vibration frequency to better overcome the elasticity of the glue.

[0076] Example: In a motor stator dispensing application, the vision system detects that the glue line after crossing has obvious defects; after analysis, the performance error is decoupled: the breakage morphology error is small, indicating that the pullback action is basically successful; but the spread morphology error is very large, indicating that the glue line starts poorly; accordingly, the system determines that the current problem is mainly caused by the poor effect of the second preset action sequence, so the incremental learning algorithm allocates 90% of the learning weight to the vibration control sub-module in this optimization cycle, and only fine-tunes the pullback action sub-module; after several production cycle iterations, the system is specifically strengthened for such "difficult to spread" conditions, so that when similar discontinuities are encountered in the future, it can directly output more optimal vibration parameters, thereby quickly eliminating such defects.

[0077] In S2, the dedicated compensation model also takes the wetting property of the substrate surface as one of the input parameters for dynamic decision making; when the model identifies the substrate as a low surface energy material, the energy level of the high-frequency micro-vibration will be dynamically increased to overcome the energy barrier when the colloid starts to spread on the rear trajectory segment.

[0078] In actual dispensing applications, the final form quality of the glue line is greatly limited by the interaction between the colloid and the substrate material; the wetting property of the substrate surface, in a simple way, describes the difficulty of the liquid colloid spreading and adhering on the solid substrate surface, which is a physical concept that integrates parameters such as surface tension and surface energy.

[0079] For high surface energy materials (such as clean metals, glass), the colloid is easy to spread, forming a low contact angle, wide and flat glue line cross-section.

[0080] For low surface energy materials (such as many plastics, fluorine-coated or anti-fouling coated substrates), the colloid will tend to "shrink" into a spherical shape, making it difficult to spread, forming a high contact angle, narrow and high glue line cross-section, and the adhesion is also poor.

[0081] In the S2 step, the model obtains the substrate wetting property parameters of the current dispensing area through the pre-stored material database or integrated surface sensing module.

[0082] When the model identifies the substrate as a low surface energy material, it determines that the colloid will face a higher energy barrier when it starts to spread on the rear trajectory segment. Here, the energy barrier is a physical and chemical term, which means the energy obstacle that the colloid molecules need to overcome to transition from the initial state to the stable spreading state on the substrate.

[0083] In order to overcome this obstacle, the model dynamically adjusts the strategy of the second preset action sequence, and the core measure is to increase the energy level of the high-frequency micro-vibration. The energy level here is a comprehensive indicator, which is usually positively correlated with the amplitude and power of the vibration. The purpose of increasing the energy level is to give the colloid molecules higher energy by more intense vibration, so that they have enough power to break through the repulsion between the colloid and the low surface energy substrate; higher vibration energy can "crack" the air film or impurities that hinder the contact between the colloid and the substrate, forcing the colloid to achieve close molecular-level contact with the substrate in an instant.

[0084] Example: Suppose a dispensing task needs to be performed on a low surface energy engineering plastic (e.g. LCP) commonly used in electronic components, which has a water contact angle greater than 90 degrees and is a typical difficult-to-wet material. If a high-frequency micro-vibration parameter (e.g. amplitude of 5 microns) for metal substrates is used, the starting end of the glue line will form an irregular ball due to the inability to spread, resulting in a failure to lap. At this time, the special compensation model will immediately calculate that the vibration energy level needs to be increased by about 30% (e.g. by increasing the amplitude to 6.5 microns) after receiving the input "substrate is LCP material"; this adjustment allows the glue to obtain additional energy to overcome its high energy barrier at the moment of contact with the substrate, thereby forming a relatively flat starting end that is in good contact with the substrate, laying a solid foundation for the subsequent successful air wire lapping or preset glue anchor lapping.

[0085] In S4, the incremental learning algorithm is responsible for fine-tuning the special compensation model to adapt to the current task, however, this is only the first layer of system learning, after which the meta-optimization sub-process is started, which is not directly optimizing the current process, but to improve the system's initial response capability when facing future unforeseen discontinuity types, i.e. fast generalization capability.

[0086] This sub-process is a systematic experience refinement process, including:

[0087] First, all the key context information of the current discontinuity is standardized, including its geometric characteristics (such as span length, corner angle), glue properties (such as viscosity, rheology), and substrate wetting characteristics, and is encoded into a unified feature vector; this vector is like the "digital fingerprint" of the process problem, uniquely defining its challenges; the system then uses this "digital fingerprint" as a query condition to retrieve a certain number (e.g. K=5) of most similar historical optimization cases from the constantly enriched process knowledge base; the next step is the core "learning" step, the algorithm will analyze this group of similar cases in depth to find a key rule: what is the mapping relationship from the initial model parameter setting to the final optimized performance peak, for example, analysis may find that for a certain type of corner with specific geometric characteristics, if the energy parameter of high-frequency micro-vibration in the initial model is preset to be higher than the conventional value by a certain percentage, the number of iterations required to converge to the optimal solution will be significantly reduced; finally, the rule obtained from the above analysis is used to optimize a globally shared model initialization strategy library; this strategy library is a more advanced, empirical "dispatch center", its role is that when the system encounters a new type of discontinuity next time, even if it has never been directly processed, the system can first generate a feature vector for it, and then use this optimized strategy library to directly output an initial model parameter set better than the default value.

[0088] Example: Suppose the system first processes a large-span right-angle turn discontinuity on a low-surface-energy plastic, after several rounds of iteration, the incremental learning algorithm finds the optimal parameters, then the meta-optimization sub-process starts, it finds that the success of this optimization is largely due to the initial appropriate boost of the vibration energy; it associates the feature combination of "low surface energy + large-span right angle" with the successful strategy of "initial high-energy vibration", strengthens and stores it in the model initialization strategy library; a week later, a large-span sharp corner on another low-surface-energy coating appears on the production line, although the specific parameters are different, but its feature vector is highly similar to the previous case in the "low wettability" and "large span" two dimensions, at this time, the system no longer starts from the general default parameters, but directly calls the strategy library to recommend an initial parameter set with a higher energy level for this new discontinuity, as a result, the system only needs 1 round of fine-tuning to achieve satisfactory results, while starting from zero may take 4 to 5 rounds.

[0089] In S3, the frequency characteristics of high-frequency micro-vibration are set within a certain range, which aims to stimulate the colloidal to produce thixotropic effect, so as to achieve faster spreading and setting at the beginning of the post-trajectory segment; thixotropic effect is the key physical principle of precise control realized by the present application, which describes the apparent viscosity (i.e. the degree of difficulty of flow) of a fluid will temporarily decrease with the increase of shear rate (such as stirring, vibration), and the viscosity will gradually recover after the shear action stops.

[0090] In the present method, by applying high-frequency micro-vibration, a high-frequency shear action is applied to the colloidal, which causes the apparent viscosity of the colloidal to decrease significantly in an instant, making it more fluid.

[0091] The frequency characteristics of high-frequency micro-vibration are set within a certain range of 100 Hz to 2000 Hz (example), which is based on the following multiple considerations:

[0092] The lower limit (such as 100 Hz) is set: too low frequency cannot effectively stimulate thixotropic effect, only when the vibration frequency is high enough, it can continuously and quickly impact the internal structure network of the colloidal, so that it cannot be rebuilt in time, thus maintaining a low viscosity and easy to spread state; below this lower limit, the shear action is discontinuous, the viscosity of the colloidal does not decrease significantly, and it is difficult to overcome the spreading energy barrier.

[0093] The upper limit (such as 2000 Hz) is set: too high frequency will cause a series of problems, first, too high frequency may cause the vibration energy to dissipate in the form of heat, rather than effectively changing the structure of the colloidal, second, it may cause the colloidal to splash or produce unnecessary standing waves, which destroys the morphological consistency of the glue line, in addition, from the perspective of engineering implementation, generating and controlling high-frequency micro-vibration is demanding on equipment and has low cost-effectiveness.

[0094] Therefore, the range of 100 Hz to 2000 Hz is generally an optimized interval that can reliably and efficiently stimulate the thixotropic effect of the glue while avoiding side effects.

[0095] At the critical moment of starting to spread in the later trajectory segment, the dispensing needle performs vibration at this specific frequency; due to the instantaneous reduction of the viscosity of the glue, its flow resistance decreases, and the spreading speed on the substrate surface is greatly improved, which helps to quickly form a stable and wide lap joint substrate.

[0096] When the needle moves away and the vibration stops, the thixotropic effect enables the viscosity of the glue to quickly recover. This rapid "thickening" feature enables the glue line to quickly fix its shape after spreading, preventing it from flowing or retracting due to its own gravity or surface tension, and ensuring the accuracy and stability of the lap joint geometry.

[0097] For example, take an epoxy resin glue used for chip underfilling as an example. This glue has significant thixotropy. When air-wire lap joint is needed, high-frequency micro-vibration with a dispensing needle opening frequency of 800 Hz and an amplitude of 10 microns is used. At the moment of contacting the substrate, the viscosity of the glue decreases sharply under the action of vibration, enabling it to spread rapidly to the target width within 50 milliseconds. Subsequently, the vibration stops, and the viscosity of the glue recovers to near the initial state within 100 milliseconds, "locking" the formed lap joint in place, perfectly avoiding the risk of collapse or bridge short circuit caused by excessive thinning of the glue.

[0098] Example Two:

[0099] As shown in Figure 2 , this embodiment further improves the design based on Example One. The difference is that in Example One, it is found that when there are multiple high-density, rapidly alternating discontinuities in the dispensing path, the system's calculation and decision load will increase dramatically in the continuous execution of the "perception-decision-compensation-learning" loop of S3 to S4, causing system response delay and physical movement of the dispensing equipment to be temporarily suspended to wait for the decision result, failing to achieve truly smooth and uninterrupted efficient production. Based on this, a dispensing control method for discontinuous paths further includes S5: establishing and maintaining a local cache of discontinuity compensation strategies.

[0100] This step is activated after S1 and runs asynchronously independent of the main control loop (S2-S4). Its specific process is as follows:

[0101] When the system is idle or computing resources are sufficient, a "virtual planning" process is pre-executed, which identifies all planned discontinuities in advance based on the complete dispensing path obtained from the upstream CAD / CAM system; for each identified discontinuity, the system simulates the core steps of S3 and S4 in advance, i.e., according to the predicted characteristics of the discontinuity, the corresponding special compensation model and initial parameters are pre-loaded or quickly generated from the process knowledge base and model initialization strategy library, and these "pre-compiled" compensation strategies are packaged into a lightweight strategy data packet and stored in a local cache of the device controller.

[0102] When the dispensing equipment runs to the discontinuity in actual production, its control logic no longer completely relies on real-time and high-load model calculation, but preferentially queries and directly calls the prepared strategy data packet from the local cache, which greatly reduces the calculation delay in the critical process window period.

[0103] At the same time, the system maintains the real-time nature of the cache, and if significant optimization updates are made to the compensation model for a certain discontinuity during the incremental learning process of S4, the system will update the corresponding strategy data packet in the cache in real time, ensuring that the latest and optimal strategy is used in subsequent production.

[0104] Embodiment Three:

[0105] As shown in Figure 3 Based on the same inventive concept as the dispensing control method for discontinuous paths in the preceding embodiments, the present application provides a dispensing control system for discontinuous paths, and the system and method embodiments in the present application are based on the same inventive concept. The system comprises:

[0106] A trajectory analysis module obtains a three-dimensional dispensing trajectory containing at least one path discontinuity, and identifies the trajectory segments before and after the path discontinuity, denoted as the front trajectory segment and the rear trajectory segment;

[0107] A behavior modeling module predicts the breaking and joining behavior of the glue at the path discontinuity based on the termination geometric characteristics of the front trajectory segment, the starting geometric characteristics of the rear trajectory segment, and the spatial span of the path discontinuity, and establishes a special compensation model for controlling the shape of the glue line at both ends of the path discontinuity;

[0108] The dynamic execution module executes a first preset action sequence according to the special compensation model to form a front glue line with a target ending shape before the dispensing head moves to the path discontinuity point along the front trajectory segment; and executes a second preset action sequence according to the special compensation model to form a starting shape that matches the ending shape of the front glue line at the starting point of the rear trajectory segment when the dispensing head crosses the path discontinuity point and is about to start the rear trajectory segment.

[0109] The online evolution module compares the ideal shape predicted by the special compensation model with the actual three-dimensional appearance of the glue line obtained through the visual detection system based on the physical data synchronously collected when the action sequence is executed, generates a performance error, fine-tunes the parameters of the special compensation model through an incremental learning algorithm using the performance error, and archives the process data of this round and the optimized model parameters to the process knowledge base.

[0110] Obviously, various modifications and changes can be made to the present application by those skilled in the art without departing from the spirit and scope of the present application. Thus, it is intended that the present application cover the modifications and changes as long as they come within the scope of the claims of the present application and their equivalents.

[0111] The above description is merely preferred embodiments of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can make equivalent replacements or changes to the technical solutions and concepts of the present application within the technical scope disclosed by the present application, and such replacements or changes should be covered within the protection scope of the present application.

Claims

1. A method for controlling dispensing of a discontinuous path, the method comprising: The method comprises: acquiring a three-dimensional dispensing trajectory containing at least one path discontinuity, and identifying trajectory segments before and after the path discontinuity, denoted as a front trajectory segment and a rear trajectory segment; based on the termination geometric feature of the front trajectory segment, the initiation geometric feature of the rear trajectory segment, and the spatial span of the path discontinuity, predicting the breaking and bridging behavior of the adhesive at the path discontinuity, and establishing a special compensation model for controlling the morphology of the adhesive thread at both ends of the path discontinuity; the special compensation model outputs a decision by comprehensively analyzing the dynamic balance relationship between the spatial span of the path discontinuity and the cohesive force and tensile viscosity of the adhesive itself, the decision including an in-air bridging mode and a substrate-reserved bridging mode; when the decision is the in-air bridging mode, the adhesive thread is controlled to cross the discontinuity in a natural parabolic trajectory affected by air resistance and gravity, and the end point of the adhesive thread overlaps with the starting area of the rear trajectory segment within a preset range; when the decision is the substrate-reserved bridging mode, the dispensing head forms an adhesive anchor point with a predetermined volume and shape at the end of the front trajectory segment before reaching the path discontinuity, and the anchor point serves as a physical basis for bridging at the starting point of the rear trajectory segment; before the dispensing head moves along the front trajectory segment to the path discontinuity, a first preset action sequence is performed according to the special compensation model to form a front adhesive thread with a target end shape; when the dispensing head crosses the path discontinuity and is about to start the rear trajectory segment, a second preset action sequence is performed according to the special compensation model to form a starting shape at the starting point of the rear trajectory segment, which is adapted to the end shape of the front adhesive thread; based on the physical data synchronously collected during the execution of the action sequences and the actual three-dimensional morphology of the adhesive thread subsequently acquired by a visual detection system, the ideal morphology predicted by the special compensation model is compared to generate a performance error, the performance error is used to fine-tune the parameters of the special compensation model through an incremental learning algorithm, and the process data of this round and the optimized model parameters are archived to a process knowledge base.

2. The method of claim 1, wherein, The operation mechanism of the special compensation model includes: taking the spatial vector of the path discontinuity, the rheological parameters of the adhesive, and the dynamic crossing speed of the dispensing head as inputs, and synchronously solving the amplitude and speed of the pullback action, and the frequency characteristics and energy level of the high-frequency micro-vibration through a built-in dynamic function representing the viscoelastic properties of the adhesive.

3. The method of claim 1, wherein, The spatial motion path followed by the dispensing head when crossing the path discontinuity is a three-dimensional spline curve dynamically generated according to the tension optimization target output by the special compensation model, which ensures the stability of the tension of the adhesive thread during the crossing process and the accuracy of the landing point.

4. The method of claim 1, wherein, The performance error is decoupled into a breaking morphology error corresponding to the end shape of the front adhesive thread and a spreading morphology error corresponding to the starting shape of the rear adhesive thread; the incremental learning algorithm selectively reinforces learning of sub-modules in the special compensation model that control the pullback action and high-frequency micro-vibration, respectively, according to the contribution degrees of the breaking morphology error and the spreading morphology error with unequal weights.

5. The method of claim 1, wherein, The special compensation model also takes the wetting characteristics of the substrate surface as one of the input parameters of dynamic decision; when the model identifies that the substrate is a low surface energy material, the energy level of high-frequency micro-vibration will be dynamically increased to overcome the energy barrier of the colloid when it starts to spread at the end of the post-trajectory segment.

6. The method of claim 1, wherein, The incremental learning algorithm will start a meta-optimization sub-process after completing the parameter fine-tuning of the current special compensation model; first, the geometric characteristics of the current path breakpoint, the colloid properties and the substrate wetting characteristics are coded into a unified feature vector; then, based on this feature vector, K most similar historical process cases are retrieved from the process knowledge base, where K is a pre-set positive integer; further, by analyzing the correlation between the initial model parameters and the final optimized performance in the cases, a globally shared model initialization strategy library is optimized; the optimized strategy library is used to directly generate a set of initial model parameters better than the default values for new breakpoints encountered subsequently.

7. The method of claim 2, wherein, The frequency characteristics of the high-frequency micro-vibration are set within a specific range, which aims to stimulate the colloidal to produce thixotropic effect, so as to achieve faster spreading and shaping at the beginning of the post-trajectory segment.

8. The method of claim 1, wherein, The method further comprises: establishing and maintaining a local cache of path breakpoint compensation strategies.

9. A glue dispensing control system for a discontinuous path, characterized by, The system comprises: a trajectory analysis module that obtains a three-dimensional dispensing trajectory containing at least one path breakpoint, and identifies the trajectory segments before and after the path breakpoint, denoted as the pre-trajectory segment and the post-trajectory segment; a behavior modeling module that predicts the breaking and joining behavior of the colloid at the path breakpoint based on the termination geometric characteristics of the pre-trajectory segment, the starting geometric characteristics of the post-trajectory segment, and the spatial span of the path breakpoint, and establishes a special compensation model for controlling the morphology of the glue line at both ends of the path breakpoint; the special compensation model outputs decisions by comprehensively analyzing the dynamic balance relationship between the spatial span of the path breakpoint and the cohesive force and tensile viscosity of the colloid, the decisions including the in-air wire joining mode and the substrate reserved joining mode; when the decision is the in-air wire joining mode, the control of the pulled-out glue wire is to cross the breakpoint in a natural parabolic trajectory affected by air resistance and gravity, and to make its end point form a preset range of overlap with the starting area of the post-trajectory segment; when the decision is the substrate reserved joining mode, the dispensing head forms a colloid anchor point with a predetermined volume and shape at the end of the pre-trajectory segment before reaching the path breakpoint, which serves as the physical basis for joining at the starting point of the post-trajectory segment; a dynamic execution module that performs a first preset action sequence according to the special compensation model to form a pre-glue line with a target end shape before the dispensing head moves along the pre-trajectory segment to the path breakpoint, and performs a second preset action sequence according to the special compensation model to form a starting shape at the starting point of the post-trajectory segment that matches the end shape of the pre-glue line when the dispensing head crosses the path breakpoint and is about to start the post-trajectory segment. An online evolution module compares the performance error generated by comparing the actual three-dimensional appearance of the bead wire obtained through visual detection system after the action sequence is performed with the ideal appearance predicted by the special compensation model, and uses the performance error to fine-tune the parameters of the special compensation model through an incremental learning algorithm, and archives the process data of this round and the optimized model parameters to the process knowledge base.

Images