CN117483996A - Automatic welding device and automatic positioning method for heating wire for electronic cigarette production - Google Patents
Automatic welding device and automatic positioning method for heating wire for electronic cigarette production Download PDFInfo
- Publication number
- CN117483996A CN117483996A CN202311582552.3A CN202311582552A CN117483996A CN 117483996 A CN117483996 A CN 117483996A CN 202311582552 A CN202311582552 A CN 202311582552A CN 117483996 A CN117483996 A CN 117483996A
- Authority
- CN
- China
- Prior art keywords
- heating wire
- wire
- model
- data
- sleeve
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000010438 heat treatment Methods 0.000 title claims abstract description 136
- 238000003466 welding Methods 0.000 title claims abstract description 50
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 45
- 239000003571 electronic cigarette Substances 0.000 title claims abstract description 32
- 230000007246 mechanism Effects 0.000 claims abstract description 62
- 238000004804 winding Methods 0.000 claims abstract description 31
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000010801 machine learning Methods 0.000 claims abstract description 15
- 238000005457 optimization Methods 0.000 claims abstract description 14
- 238000013136 deep learning model Methods 0.000 claims abstract description 11
- 238000003825 pressing Methods 0.000 claims description 44
- 230000006870 function Effects 0.000 claims description 43
- 238000000605 extraction Methods 0.000 claims description 20
- 239000011159 matrix material Substances 0.000 claims description 18
- 238000000513 principal component analysis Methods 0.000 claims description 18
- 238000009826 distribution Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 14
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 10
- 230000004913 activation Effects 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000013178 mathematical model Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 7
- 238000011217 control strategy Methods 0.000 claims description 6
- 238000002156 mixing Methods 0.000 claims description 6
- 239000002131 composite material Substances 0.000 claims description 5
- 239000000203 mixture Substances 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 230000004044 response Effects 0.000 claims description 5
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 238000003709 image segmentation Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 238000005192 partition Methods 0.000 claims description 4
- 230000000149 penetrating effect Effects 0.000 claims description 4
- 238000000611 regression analysis Methods 0.000 claims description 4
- 238000013519 translation Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 3
- 230000008901 benefit Effects 0.000 claims description 2
- 230000004069 differentiation Effects 0.000 claims description 2
- 238000003708 edge detection Methods 0.000 claims description 2
- 230000010355 oscillation Effects 0.000 claims description 2
- 230000009467 reduction Effects 0.000 claims description 2
- 238000006467 substitution reaction Methods 0.000 claims description 2
- 238000012423 maintenance Methods 0.000 abstract description 3
- 238000013461 design Methods 0.000 abstract description 2
- 230000009471 action Effects 0.000 description 9
- 239000000047 product Substances 0.000 description 9
- XKRFYHLGVUSROY-UHFFFAOYSA-N Argon Chemical compound [Ar] XKRFYHLGVUSROY-UHFFFAOYSA-N 0.000 description 6
- 238000007599 discharging Methods 0.000 description 4
- 229910052786 argon Inorganic materials 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000007774 longterm Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000001125 extrusion Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 239000010953 base metal Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 239000012467 final product Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K31/00—Processes relevant to this subclass, specially adapted for particular articles or purposes, but not covered by only one of the preceding main groups
- B23K31/02—Processes relevant to this subclass, specially adapted for particular articles or purposes, but not covered by only one of the preceding main groups relating to soldering or welding
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K37/00—Auxiliary devices or processes, not specially adapted to a procedure covered by only one of the preceding main groups
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Mechanical Engineering (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Optics & Photonics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Wire Processing (AREA)
Abstract
The invention discloses an automatic welding device and an automatic positioning method for a heating wire for electronic cigarette production, which belong to the field of welding and machine learning, and the winding mechanism is adopted to enable a wire to be uniformly wound on the end part of a pin of the heating wire, so that the welding quality of the wire and the heating wire is improved; the positioning precision of the heating wire is obviously improved through advanced image processing and a deep learning model; the model can adapt to the changes of different heating wires and semicircular grooves, and the adaptability of the system is improved; accurate positioning and feedback control reduce errors in the production process and improve the product quality; the optimization algorithm and the rapid model training reduce the adjustment time of the production line and improve the efficiency; the integrated system reduces the dependence on complex mechanical adjustment and reduces the maintenance cost; the scheme design allows the problem of other automatic production lines to be easily adapted, and has good expansibility.
Description
Technical Field
The invention belongs to the technical field of heating wire welding and machine learning, and particularly relates to an automatic heating wire welding device for electronic cigarette production and an automatic heating wire positioning method.
Background
Welding of heating wires for electronic cigarette production is a key link in an electronic cigarette manufacturing process, and the conventional process mostly adopts an artificial resistance welding mode to finish welding connection of heating wires and wires;
chinese patent publication No. CN116441783a discloses an automatic welding device for heating wire for electronic cigarette production, which adopts a mechanical arm to place the heating wire into a positioning mechanism; winding the wire on the heating wire by using the movable pressing strip and the rotating frame; finally, the connection is completed by pressure welding, the connection strength between the wire and the heating wire is increased by winding the wire, but the device has the following defects: the winding mode is simple, and wire winding is not neat and complete enough, and further connection strength can be influenced.
In order to solve the problems, an automatic welding device for the heating wire for the electronic cigarette production, which is accurate in winding and good in welding quality, needs to be developed.
The current positioning depends on mechanical processing and a simple clamp, and has limited adjustability, so that the heating wire cannot be accurately positioned at an ideal position, and eccentric deviation in the subsequent winding process is caused. The connection strength of the wires and the heating wires can be directly weakened, and the product quality is affected. Therefore, a positioning model needs to be trained by machine learning, a semicircular groove image can be analyzed, geometric features of the heating wire are extracted, and a coordinate mapping relation is modeled and learned, so that high-precision and automatic heating wire positioning is realized.
Disclosure of Invention
The embodiment of the invention provides an automatic welding device and an automatic positioning method for a heating wire for electronic cigarette production, which are used for solving the problems in the prior art.
The embodiment of the invention adopts the following technical scheme: the automatic welding device for the heating wire for producing the electronic cigarette comprises a processing table, a welding mechanism and a winding mechanism, wherein the winding mechanism is used for winding a wire on the pin end of the heating wire;
the winding mechanism includes:
the rotating frame is arranged on the processing table, and a rotating hole is transversely formed in the rotating frame;
the rotary table is arranged in the rotary hole in a rotary way, an eccentric hole is formed in the rotary table, and the eccentric hole is not coaxial with the rotary table;
the sleeve is arranged on the left side of the rotating disc, is fixedly connected with the rotating disc and is coaxial with the eccentric hole;
the first pressing mechanism is arranged on the sleeve and used for pressing the right end part of the wire;
the first platform is arranged on the processing platform, the upper end of the first platform is provided with a semicircular groove for accommodating the pin end of the heating wire, the semicircular groove is coaxial with the turntable, and the first platform is provided with a second pressing mechanism for clamping the left end part of the wire and the pin end of the heating wire together;
and a rotary driving mechanism for driving the turntable to rotate.
Through adopting winding mechanism, drive the wire at the sleeve and rotate for the axis around the heater strip pin end after fixed, under the both ends of wire all by fixed circumstances to can make the even winding of wire at heater strip pin end, and improved the welding quality of wire and heater strip.
Further, the first pressing mechanism includes:
the sliding sleeve is elastically and slidably arranged in the sleeve, the top of the sliding sleeve is provided with an arc-shaped groove longitudinally extending to the inside of the sliding sleeve, and an arc-shaped block matched with the arc-shaped groove is arranged in the arc-shaped groove;
and a first linear driving source arranged above the sleeve, wherein a sliding groove extending along the length direction of the sliding groove is formed in the outer wall of the upper part of the sleeve, a sliding block is connected in the sliding groove in a sliding manner, a strip-shaped avoidance groove penetrating into the sleeve is formed in the bottom of the sliding groove, the length direction of the strip-shaped avoidance groove is consistent with the length direction of the sliding groove, the first linear driving source is arranged on the sliding block, and an output shaft of the first linear driving source penetrates through the sliding block and penetrates through the avoidance groove to be fixedly connected with the arc-shaped block.
The right end part of the wire is fixed under the action of the sliding sleeve, and a certain movable space is provided for the right end part of the wire during winding. A baffle plate is arranged at the left end part of the sleeve, and a round hole which is coaxial with the eccentric hole is formed in the baffle plate;
The inside coaxial spring that is provided with of sleeve, the both ends of spring are respectively elasticity and are contradicted on baffle and sliding sleeve.
Under the elastic action of the spring, the wire can be always in a tight state, so that the wire can be uniformly wound on the pin end of the heating wire.
The outer wall department of sliding sleeve is provided with the stopper along its axial extension, sleeve inner wall department is provided with the spacing groove along its axial extension, the stopper slides and sets up in the spacing groove.
The rotation driving mechanism includes:
the outer gear ring is fixedly sleeved on the turntable, and the outer gear ring is positioned at the left side of the rotating frame;
the main gear is arranged on the left side of the rotating frame and is in meshed connection with the outer gear ring;
and the output shaft of the rotary driving source penetrates through the rotating frame and is coaxially and fixedly connected with the main gear.
Further, the second pressing mechanism comprises two pressing plates, the two pressing plates are arranged in a mirror image mode on the axis of the semicircular groove, and the lower ends of the two pressing plates are located at the same height with the top surface of the semicircular groove;
the outside of two clamp plates all is provided with the motor frame, all is provided with the second sharp actuating source in every motor frame, and the output shaft of every second sharp actuating source all fixed connection is on corresponding clamp plate.
Further, the opposite ends of the tops of the two pressing plates are provided with an inner arc plate, and the diameters of the two inner arc plates are matched with the diameter of the lead.
The automatic positioning method of the heating wire of the automatic welding device for the heating wire for producing the electronic cigarette comprises the following steps:
step 1: image processing and feature extraction
1.1. Image acquisition and pretreatment:
1.11. acquiring images of the heating wire and the semicircular groove by using a high-resolution industrial camera;
1.12. segmenting the image by using a Gaussian Mixture Model (GMM), and highlighting key areas of the heating wire and the semicircular groove (41);
1.13. image enhancement using wavelet transform to improve sharpness of edge features;
1.2. feature extraction and optimization: extracting key features from the enhanced image by using a method combining Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), and constructing a feature vector quantity F for subsequent coordinate regression analysis through the features;
step 2, complex coordinate regression model
2.1. Deep learning model construction: designing a Deep Neural Network (DNN) comprising a convolution layer, a batch normalization layer, a ReLU activation layer and a full connection layer, wherein the target of the deep neural network is to predict the ideal position coordinates of the heating wire from the characteristic vector F;
2.2. Coordinate regression formula: defining model outputs as delta X and delta Y, and representing deviation between an ideal position and an actual position of the heating wire;
coordinate regression is represented using a composite function:
ΔX＝f(W 1 ·h(F)+b 1 )
ΔY＝g(W 2 ·h(F)+b 2 )
wherein f and g are nonlinear mapping functions, h is a feature extraction function, W 1 ,W 2 ,b 1 ,b 2 Is a network parameter;
step 3: model training and optimization
3.1. Loss function and optimization:
using a weighted combined loss function, combining Mean Square Error (MSE) and cross entropy loss to optimize positioning accuracy and stability; then, an Adam optimizer is applied to combine the momentum and the self-adaptive learning rate strategy so as to accelerate the convergence speed and improve the stability of the model;
step 4, system integration and cooperative control
The deviation value output by the model is used for adjusting the movement strategy of the manipulator, so that closed-loop control is realized; combines a machine learning model and a traditional PID control algorithm to realize more accurate dynamic adjustment.
The above at least one technical scheme adopted by the embodiment of the invention can achieve the following beneficial effects:
1. according to the invention, the winding mechanism is adopted, the sleeve drives the wire to rotate around the fixed heating wire pin end as the axis, and under the condition that both end parts of the wire are fixed, the wire can be uniformly wound on the heating wire pin end part, and the welding quality of the wire and the heating wire is improved.
2. According to the invention, under the elastic action of the spring, the wire can be always in a tight state, so that the wire can be uniformly wound on the pin end of the heating wire, the right end part of the wire cannot rotate around the pin end of the heating wire in a tight state if the right end part of the wire is unfixed in the winding process, the wire cannot be uniformly wound on the pin end of the heating wire, and the wire cannot be wound if the right end part of the wire is fixed and cannot move, so that after the wire is fixed with the wire through the sliding sleeve, under the elastic action of the spring, the wire can be always in the tight state, and a certain contraction space is provided for the right end part of the wire when the wire is wound, so that the winding uniformity of the wire is ensured, and the connection reliability between the wire and the heating wire is improved.
3. Positioning accuracy is improved: the invention can more accurately identify and position the position of the heating wire through advanced image processing and feature extraction technologies such as Gaussian Mixture Model (GMM) and wavelet transformation, and combination of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA); this accurate feature extraction directly improves the accuracy of the positioning.
4. Generalization ability of enhancement model: by utilizing the deep learning model, particularly the strong characteristic learning capability of a Convolutional Neural Network (CNN), the invention can process various complex and variable production environments and improve the adaptability of the model to different heating wires and semicircular grooves.
5. Production errors are reduced, and product quality is improved: through an accurate coordinate regression model and a feedback control mechanism, the invention can effectively reduce the deviation of the positioning of the heating wire, thereby reducing the error in the production process and improving the quality of the final product.
6. The production efficiency is improved: the integrated system and the cooperative control strategy are combined with a machine learning model and a PID control algorithm, so that the whole production process is more automatic and efficient; this increase in automation can significantly increase the throughput of the production line.
7. Enhancing the stability and reliability of the system: by using the weighted combination loss function and the Adam optimizer, the invention realizes more stable and reliable performance in the model training process; this stability is critical for a production line running for a long period of time.
8. Adaptability and flexibility: through the self-learning capability of the machine learning model, the invention can adapt to small changes on the production line and different production conditions, and provides greater flexibility.
9. The long-term operation cost is reduced: the invention is helpful to reduce the rejection rate and the maintenance cost by reducing the production error and improving the product quality, thereby reducing the long-term operation cost. The data collection and analysis capabilities of the present invention can provide insight into the production process and help to continuously improve and optimize the production process. In conclusion, the method not only improves the efficiency and quality of the automatic production line of the electronic cigarette, but also provides higher flexibility and reliability for the production process, and simultaneously reduces the long-term operation cost.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention and do not constitute a limitation on the invention. In the drawings:
FIG. 1 is a schematic perspective view of the present invention;
FIG. 2 is a schematic perspective view of a winding mechanism according to the present invention;
FIG. 3 is a perspective cross-sectional view of the winding mechanism of the present invention;
FIG. 4 is an enlarged view of FIG. 3 at A;
FIG. 5 is an exploded view of a partial structure of the winding mechanism of the present invention;
FIG. 6 is a partial exploded view of the turret, turntable, and rotary drive mechanism of the present invention;
FIG. 7 is a schematic perspective view of a first platform and a second hold-down mechanism according to the present invention;
fig. 8 is a flowchart of the heating wire automatic positioning method of the present invention.
Reference numerals
1-a processing table; 2-winding mechanism; 21-a rotating frame; 211-a rotation hole; 22-a turntable; 221-eccentric holes; 23-sleeve; 231-a sliding groove; 232-a strip-shaped avoidance groove; 233-a limit groove; 3-a first hold-down mechanism; 31-sliding sleeve; 311-arc grooves; 312-limiting blocks; 32-arc blocks; 33-a first linear drive source; 34-a slider; 35-a separator; 351-round holes; 36-a spring; 4-a first platform; 41-a semicircular groove; 5-a second pressing mechanism; 51-pressing plate; 511-inner arcuate plate; 52-a motor frame; 53-a second linear drive source; 6-a rotary drive mechanism; 61-an outer gear ring; 62-a main gear; 63-a rotary drive source; 7-heating wire pin ends; 8-conducting wires.
Detailed Description
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description of the applications herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having" and any variations thereof in the description and claims of the present application and in the description of the figures above are intended to cover non-exclusive inclusions. The terms first, second and the like in the description and in the claims or in the above-described figures, are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.
Example 1
Referring to fig. 1 to 7, an embodiment of the present invention provides an automatic welding device for a heating wire for electronic cigarette production, which includes a processing table 1, a welding mechanism, and a winding mechanism 2 for winding a wire 8 around a pin end 7 of the heating wire; the winding mechanism 2 includes: a rotating frame 21, wherein the rotating frame 21 is arranged on the processing table 1, and a rotating hole 211 is transversely arranged on the rotating frame 21; rotating the rotary table 22 arranged in the rotating hole 211, wherein an eccentric hole 221 is formed in the rotary table 22, and the eccentric hole 221 is not coaxial with the rotary table 22; a sleeve 23 arranged on the left side of the turntable 22, wherein the sleeve 23 is fixedly connected with the turntable 22, and the sleeve 23 is coaxial with the eccentric hole 221; the first pressing mechanism 3 is arranged on the sleeve 23 and used for pressing the right end part of the wire 8; the first platform 4 is arranged on the processing platform 1, a semicircular groove 41 for accommodating the heating wire pin end 7 is formed in the upper end of the first platform 4, the semicircular groove 41 is coaxial with the turntable 22, and the first platform 4 is provided with a second pressing mechanism 5 for clamping the left end part of the wire 8 and the heating wire pin end 7 together; and a rotation driving mechanism 6 for driving the turntable 22 to rotate.
In this embodiment, the automatic welding device for heating wires for electronic cigarette production further includes an automatic feeding and discharging mechanism for feeding and discharging the heating wires, where the automatic feeding and discharging mechanism is a manipulator (the manipulator is in the prior art and is not shown in the drawing), an executing part of the manipulator is provided with a fixture adapted to the heating wires, and the manipulator moves to a specified position to clamp the heating wires and place the pin ends thereof into the semicircular grooves 41;
after the heating wire is welded, a finished product is taken out by a manipulator, so that automatic feeding and discharging of the heating wire are realized, the welding mechanism is in the prior art and is not shown in the drawing, and note that in order to improve the welding quality, the welding mechanism adopts an argon arc welding head to replace the original welding resistance mode, so that the welding quality is improved, an argon arc is generated through a welding torch, a base metal is melted, an argon protection molten pool is melted, and high-quality welding is realized;
during processing, the mechanical arm places the pin end of the heating wire in the semicircular groove 41, the whole heating wire is made of rigid materials so as to avoid the influence of easy deformation, then one end of the wire 8 is inserted into the sleeve 23 from the eccentric hole 221, the wire 8 extends out of the left end part of the sleeve 23, the wire 8 is right above the pin end 7 of the heating wire, after the wire 8 is overlapped with the pin end 7 of the heating wire in a proper position, the left end part of the wire 8 and the pin end 7 of the heating wire are clamped together by the second pressing mechanism 5, and then the right end part of the wire 8 is pressed through the first pressing mechanism 3, so that the whole fixing of the wire 8 is completed;
Then, the rotary table 22 is driven to rotate through the operation of the rotary driving mechanism 6, so that the sleeve 23 rotates along with the rotary table, under the condition that the left end part of the wire 8 is fixed, the wire 8 rotates around the pin end of the heating wire as an axis, so that the left end part of the wire 8 can be wound on the pin end of the heating wire, and after the wire 8 is wound and fixed on the heating wire, the welding between the wire 8 and the heating wire can be completed through the welding mechanism;
finally, after the first pressing mechanism 3 and the second pressing mechanism 5 are reset, the heating wire is stretched by the manipulator, namely the wire 8 can be completely taken out from the sleeve 23, so that the material taking of the welded heating wire and the wire 8 is completed;
through adopting winding mechanism 2, drive wire 8 at sleeve 23 and rotate for the axis around the heater strip pin end 7 after fixed, under the condition that the both ends of wire 8 are all fixed to can make wire 8 even winding at heater strip pin end 7 portion, and improve the welding quality of wire 8 and heater strip.
The first pressing mechanism 3 includes: the sliding sleeve 31 is elastically and slidably arranged in the sleeve 23, a limiting block 312 extending along the axial direction of the sliding sleeve 31 is arranged at the outer wall of the sliding sleeve 31, a limiting groove 233 extending along the axial direction of the sliding sleeve is arranged at the inner wall of the sleeve 23, the limiting block 312 is slidably arranged in the limiting groove 233, a partition plate 35 is arranged at the left end part of the sleeve 23, and a round hole 351 coaxial with the eccentric hole 221 is formed in the partition plate 35; a spring 36 is coaxially arranged in the sleeve 23, and two ends of the spring 36 elastically abut against the partition 35 and the sliding sleeve 31 respectively; the top of the sliding sleeve 31 is provided with an arc-shaped groove 311 extending longitudinally to the inside, and an arc-shaped block 32 matched with the arc-shaped groove 311 is arranged in the arc-shaped groove 311;
The first linear driving source 33 is arranged above the sleeve 23, a sliding groove 231 extending along the length direction of the sliding groove 231 is formed in the outer wall of the upper part of the sleeve 23, a sliding block 34 is connected in a sliding manner in the sliding groove 231, a strip-shaped avoiding groove 232 penetrating into the sleeve 23 is formed in the bottom of the sliding groove 231, and a certain movable space is provided for the first linear driving source 33 through the strip-shaped avoiding groove 232 and the sliding block 34;
the length direction of the strip-shaped avoidance groove 232 is consistent with the length direction of the sliding groove 231, the first linear driving source 33 is installed on the sliding block 34, and an output shaft of the first linear driving source 33 penetrates through the sliding block 34 and fixedly connected with the arc-shaped block 32 through the avoidance groove;
in this embodiment, under the elastic action of the spring 36, the sliding sleeve 31 is located at the right end of the sleeve 23, when the wire 8 is positioned, the wire 8 sequentially passes through the eccentric hole 221, the sliding sleeve 31 and the round hole 351, so that the wire 8 overlaps the heating wire pin end 7 by a certain distance, and the left end of the wire 8 is located right above the heating wire pin end 7, the first linear driving source 33 is preferably a linear cylinder, so that the first linear driver operates to tightly press the arc block 32 on the wire 8 by the compression of air pressure, and the right end of the wire 8 can be stably connected with the sliding sleeve 31;
After the second compressing mechanism 5 compresses the left end part of the wire 8 and the heating wire pin end 7 together, the rotary driving mechanism 6 drives the turntable 22 and the sleeve 23 to rotate together, so that the wire 8 can rotate around the heating wire pin end 7 as an axis, and since the left end part of the wire 8 is fixed and the right end part of the wire 8 can elastically slide in the sleeve 23 together along with the sliding sleeve 31, the wire 8 can be wound on the heating wire pin end 7, and after the wire 8 is wound on the heating wire, the whole length of the wire 8 can be shortened;
in the process, the right end of the wire 8 drives the sliding sleeve 31 to slide to the left in the sleeve 23, the spring 36 is compressed, and the limiting block 312 is arranged on the sliding sleeve 31, and the limiting groove 233 which is in sliding fit with the limiting block 312 is arranged on the sleeve 23, so that the sliding sleeve 31 is arranged in the sleeve 23 in a sliding manner, and relative rotation between the sliding sleeve 31 and the sleeve 23 is avoided;
under the elastic action of the spring 36, the wire 8 can be always in a tight state, so that the wire 8 can be uniformly wound on the heating wire pin end 7, and the wire 8 cannot rotate around the heating wire pin end 7 in a tight state due to the fact that the right end part of the wire 8 is not fixed in the winding process of the wire 8, so that the wire 8 cannot be uniformly wound on the heating wire pin end 7;
If the right end portion of the wire 8 is fixed and cannot move, the wire 8 cannot be wound, and therefore after the wire 8 is fixed through the sliding sleeve 31, the wire 8 is not only guaranteed to be always in a tight state under the elastic action of the spring 36, but also a certain shrinkage space is provided for the right end portion of the wire 8 when the wire 8 is wound, so that the winding uniformity of the wire 8 is guaranteed, and the connection reliability between the wire 8 and the heating wire is improved.
The rotation driving mechanism 6 includes: an outer gear ring 61 fixedly sleeved on the turntable 22, the outer gear ring 61 being positioned on the left side of the turret 21; a main gear 62 provided on the left side of the turret 21, the main gear 62 being engaged with the outer gear ring 61; and a rotary drive source 63 fixedly installed on the right side of the turntable 22, an output shaft of the rotary drive source 63 penetrating the turret 21 and being fixedly connected coaxially with the main gear 62;
in the present embodiment, the rotation driving source 63 operates to rotate the main gear 62, and the outer gear 61 is fixedly connected to the turntable 22 due to the meshed connection of the main gear 62 and the outer gear 61, so that the turntable 22 rotates on the rotating frame 21.
The second pressing mechanism 5 comprises two pressing plates 51, the two pressing plates 51 are arranged in a mirror image mode on the axis of the semicircular groove 41, and the lower ends of the two pressing plates 51 are located at the same height with the top surface of the semicircular groove 41; the outer sides of the two pressing plates 51 are respectively provided with a motor frame 52, each motor frame 52 is provided with a second linear driving source 53, and the output shaft of each second linear driving source 53 is fixedly connected to the corresponding pressing plate 51; the top opposite ends of the two pressing plates 51 are provided with an inner arc plate 511, and the diameters of the two inner arc plates 511 are matched with the diameters of the wires 8.
In this embodiment, after the wire 8 is fixed, the left end of the wire 8 is located between the two pressing plates 51, as shown in the figure, and then the two pressing plates 51 can be made to approach each other by the synchronous operation of the two second linear driving sources 53, the positioning of the heating wire pin end 7 is achieved under the action of the semicircular groove 41, then the positioning of the wire 8 can be achieved under the extrusion of the inner arc shape on the two pressing plates 51, and the pin end of the heating wire and the wire 8 can be fixed together under the extrusion action of the two pressing plates 51.
The process of the automatic welding device for the heating wire for producing the electronic cigarette comprises the following steps of:
s1: the manipulator coaxially places the heating wire pin end 7 in the semicircular groove 41, and simultaneously, a worker sequentially passes the lead 8 through the eccentric hole 221, the sliding sleeve 31 and the round hole 351, so that the lead 8 is positioned right above the heating wire pin end 7 and between the two pressing plates 51;
s2: the first linear driving source 33 operates, the arc-shaped block 32 compresses the lead 8 in the sliding sleeve 31, the second linear driver operates, and the two pressing plates 51 are close to each other to compress the left end part of the lead 8 and the pin end 7 of the heating wire together;
s3: the rotary driving source 63 operates, the turntable 22 and the sleeve 23 rotate together, and the left end part of the wire 8 is uniformly wound on the heating wire pin end 7;
S4: after the welding mechanism finishes the welding of the connection part of the wire 8 and the heating wire, all the mechanisms are reset, and then the wire 8 and the heating wire after the welding is finished are taken down by the manipulator.
Example 2
Referring to fig. 8, the automatic positioning method of the heating wire of the automatic welding device for producing heating wires for electronic cigarettes in embodiment 1 includes the following steps:
step 1: image processing and feature extraction
1.1. Image acquisition and pretreatment:
1.11. acquiring images of the heating wire and the semicircular groove by using a high-resolution industrial camera; a high-resolution industrial camera (CCD camera) is positioned on the processing table (1) and is arranged above the heating wire and the semicircular groove;
1.12. segmenting the image by using a Gaussian Mixture Model (GMM), and highlighting key areas of the heating wire and the semicircular groove (41);
1.13. image enhancement using wavelet transform to improve sharpness of edge features;
1.2. feature extraction and optimization: extracting key features from the enhanced image by using a method combining Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), and constructing a feature vector quantity F for subsequent coordinate regression analysis through the features;
step 2, complex coordinate regression model
2.1. Deep learning model construction: designing a Deep Neural Network (DNN) comprising a convolution layer, a batch normalization layer, a ReLU activation layer and a full connection layer, wherein the target of the deep neural network is to predict the ideal position coordinates of the heating wire from the characteristic vector F;
2.2. Coordinate regression formula: defining model outputs as delta X and delta Y, and representing deviation between an ideal position and an actual position of the heating wire;
coordinate regression is represented using a composite function:
ΔX＝f(W 1 ·h(F)+b 1 )
ΔY＝g(W 2 ·h(F)+b 2 )
wherein f and g are nonlinear mapping functions, h is a feature extraction function, W 1 ,W 2 ,b 1 ,b 2 Is a network parameter;
step 3: model training and optimization
3.1. Loss function and optimization:
using a weighted combined loss function, combining Mean Square Error (MSE) and cross entropy loss to optimize positioning accuracy and stability; then, an Adam optimizer is applied to combine the momentum and the self-adaptive learning rate strategy so as to accelerate the convergence speed and improve the stability of the model;
step 4, system integration and cooperative control
The deviation value output by the model is used for adjusting the movement strategy of the manipulator, so that closed-loop control is realized; combines a machine learning model and a traditional PID control algorithm to realize more accurate dynamic adjustment.
Data preparation and preprocessing: the deep learning model constructed in step 2 requires a large amount of training data to learn how to accurately predict the position of the heating wire. These data come from the image processing and feature extraction stage in step 1. Therefore, it is important to ensure the quality and diversity of data.
Model structure and parameter initialization: before the training process of step 3 begins, initial values of the structure (e.g., number of layers, number of neurons per layer, etc.) and parameters (e.g., weights and deviations) of the deep learning model constructed in step 2 need to be determined.
Selection of a loss function: the weighted combination loss function in step 3 directly affects the learning effect of the model in step 2. The mean square error portion focuses on reducing the error of the position prediction, while the cross entropy loss portion helps to improve classification performance (e.g., distinguishing heater wires from half slots).
Application of optimizer: the Adam optimizer is used in step 3 to adjust the parameters of the model in step 2. It updates the network parameters by calculating the gradient of the loss function, thereby reducing the prediction error.
Feedback loop: during the training process, the performance of the model needs to be continuously evaluated. If the model is found to perform poorly in predicting the heater wire position, it may be necessary to go back to step 2 to adjust the network structure or parameters, or to go back to step 1 to improve the feature extraction method.
Model verification and adjustment: after training is complete, the model needs to be tested on unseen data to verify its generalization ability. Depending on the test results, it may be necessary to readjust the model structure or training strategy.
In this way, steps 2 and 3 form an iterative loop that is continually optimized and tuned until the desired accuracy and stability is achieved. The tight connection ensures that the whole system can efficiently and accurately position the heating wire, thereby improving the automation level and the product quality of the electronic cigarette production line.
Application of model output: the model output trained in step 3 (positional deviation of the heater wire) is a key part of the PID controller input in step 4. These deviation values directly affect the tuning strategy of the PID controller.
Establishment of feedback control mechanism: in step 4, the PID controller adjusts the movement of the robot arm according to the deviation values obtained from step 3 to minimize these deviations. The feedback mechanism ensures that the system can be dynamically adjusted according to actual conditions, and improves the precision and the response speed.
Evaluation and optimization of system performance: in actual operation, the feedback control effect of step 4 needs to be continuously evaluated. If the control effect is found to be undesirable, it may be necessary to revert to step 3 to adjust the training of the model, e.g. to optimize the loss function or to adjust the optimizer parameters.
Adjustment of cooperative control strategy: the PID control parameters (proportional, integral, derivative gains) of step 4 may need to be adjusted according to the output characteristics of the model in step 3 to ensure coordinated operation of the entire system.
Perfecting a closed-loop control system: through continuous testing and tuning, steps 3 and 4 form a closed loop control system in which the output of the model directly affects the control strategy, which in turn provides feedback on the model's performance.
Utilization of real-time data: in practice, the system may collect real-time data that may be used to further optimize the model training in step 3 to more closely adapt to the actual production environment.
In this way, step 3 and step 4 form a loop that promotes each other, constantly optimizing and adjusting to achieve optimal positioning accuracy and system stability. The tight connection ensures that the whole system can efficiently and accurately position the heating wire, thereby improving the automation level and the product quality of the electronic cigarette production line.
The positioning precision of the heating wire is obviously improved through advanced image processing and a deep learning model; the model can adapt to the changes of different heating wires and semicircular grooves, and the adaptability of the system is improved; accurate positioning and feedback control reduce errors in the production process and improve the product quality; the optimization algorithm and the rapid model training reduce the adjustment time of the production line and improve the efficiency; the integrated system reduces the dependence on complex mechanical adjustment and reduces the maintenance cost; the scheme design allows the problem of other automatic production lines to be easily adapted, and has good expansibility.
The comprehensive solution combines advanced image processing technology, deep learning and feedback control mechanism to form a highly coordinated and logic system, and effectively solves the problem of positioning deviation of the heating wire.
A Gaussian Mixture Model (GMM) is a probabilistic model that assumes that all data points are mixed by a finite number of gaussian distributions. Each gaussian distribution is called a "component," and each component corresponds to a population in the data.
1.12. The GMM is used for image segmentation, and the heating wire and the semicircular groove are separated from the background; the mathematical expression of the mathematical model GMM is:
wherein: p (x) is the probability of a given data point x;
k is the number of Gaussian distributions;
φ i is the mixing coefficient of the ith Gaussian distribution and meets
Is a gaussian (normal) distribution, where mu i Is the mean value, sigma i Is a covariance matrix;
first, initializing: randomly selecting parameters (mean, covariance) and mixing coefficients of K Gaussian distributions; desired step (E-step): calculating the probability that each data point belongs to each Gaussian distribution;
maximization step (M-step): updating parameters and mixing coefficients of the gaussian distribution to maximize likelihood functions of the data;
and (3) carrying out substitution: repeating the E-step and the M-step until convergence (i.e., the variation of the parameters is very small or a preset number of iterations is reached);
input: image data, considered as multi-dimensional data points;
and (3) outputting: probability that each data point (pixel) belongs to each gaussian distribution is used for image segmentation;
1.13. The wavelet transformation is used for enhancing image features, particularly edges, so as to facilitate subsequent feature extraction and analysis;
the mathematical expression of the Continuous Wavelet Transform (CWT) is:
wherein x (t) is a signal;
ψ (t) is the wavelet function;
τ is a translation parameter;
s is a scale parameter;
ψ * (t) is the complex conjugate of the wavelet function;
the method comprises the following steps:
1. selecting a wavelet base: selecting a proper wavelet function;
2. calculating wavelet coefficients: for each scale and translation, computing an inner product of the signal and the wavelet basis;
3. reconstructing the signal: an optional step of reconstructing the original signal using the wavelet coefficients;
input image data, regarded as a two-dimensional signal; wavelet coefficients are output representing features of the image at different scales and locations, which coefficients can be used for image enhancement, particularly in terms of edge detection and feature extraction. The two methods are combined, so that useful information can be effectively extracted from complex image data, and support is provided for accurate positioning of the heating wire.
Principal Component Analysis (PCA) is a statistical method for converting a set of possibly related variables into a set of linearly uncorrelated variables, called principal components, by an orthogonal transformation. The purpose of PCA is to reduce the dimensionality of the data while retaining the most variability.
In 1.2, PCA is used to extract key features from the enhanced image and reduce the dimensionality of the data. LDA further processes these features to maximize the degree of discrimination between the different categories (heater wire and half-slot); by combining PCA and LDA, useful information can be effectively extracted from complex image data, and an optimized feature vector F is constructed for subsequent coordinate regression analysis. The method improves the efficiency and accuracy of feature extraction, and provides a solid foundation for accurate heating wire positioning.
Mathematical model the mathematical model of the mathematical model PCA is based on a covariance matrix or Singular Value Decomposition (SVD) of the data; let X be an n X p data matrix, where n is the number of samples and p is the number of variables; PCA finds a transformation matrix W such that y=xw is a new data representation, with the columns of Y being the principal components;
the method comprises the following steps:
1. normalized data: centering and scaling the original data;
2. calculating a covariance matrix:
3. solving eigenvalues and eigenvectors: performing feature decomposition on the covariance matrix;
4. and selecting main components: selecting the first k eigenvectors according to the magnitude of the eigenvalue to form a transformation matrix W;
inputting enhanced image data, which is regarded as multidimensional data points; outputting projection of data on a main component, namely data representation after dimension reduction;
Linear Discriminant Analysis (LDA) is a method of supervised learning to find feature subspaces that optimally distinguish two or more categories.
LDA finds a linear combination W to maximize the inter-class separability while minimizing the intra-class separability; this can be achieved by maximizing the following ratio:
wherein S is B Is an inter-class scatter matrix, S W Is an intra-class scatter matrix;
the specific process is as follows:
1 calculating intra-class and inter-class scatter matrices: computing S based on class labels W And S is B ；
2 solving a generalized eigenvalue problem: finding the W that maximizes J (W);
3. projection data: projecting the data onto the found feature vector;
inputting enhanced image data, which is regarded as multidimensional data points; the projection of the output data onto the LDA feature vector, i.e. the differentiation of new feature representations of different classes.
Deep learning models, particularly Deep Neural Networks (DNNs), are based on the concept of artificial neural networks. They learn and make decisions by way of modeling the human brain processing information. DNN is composed of multiple layers, each layer containing multiple neurons connected by weights and offsets.
In 2.1, the DNN is represented as a composite of a series of functions, each corresponding to a layer in the network:
Y＝f n (…f 2 (f 1 (X，W 1 ，b 1 )，W 2 ，b 2 )…，W n ，b n )
Wherein: x is input data; y is output data; f (f) i Is the activation function of the i-th layer;
W i and b i The weight and deviation of the i-th layer are respectively; the method comprises the following steps:
1. convolution layer: extracting local features of the image using a convolution operation;
2. batch normalization layer: normalizing each small batch of inputs to improve training stability and speed;
relu activation layer: applying a nonlinear activation function ReLU (x) =max (0, x);
4. full tie layer: converting the output of the previous layer into a final output format;
inputting a feature vector F, wherein the feature vector F is obtained from the steps of image processing and feature extraction; outputting deviations delta X and delta Y of the ideal position and the actual position of the heating wire;
2.2, the purpose of the coordinate regression model is to accurately predict the deviation between the ideal position and the actual position of the heating wire; by training the model, a machine learning algorithm can learn to identify and correct the position deviation of the heating wire from the image characteristics, so that the precision and the efficiency of an automatic production line of the electronic cigarette are improved.
In machine learning, the loss function is a function that measures the difference between the model predicted value and the true value. By minimizing this loss function, the performance of the model can be optimized. In some complex tasks, models can be trained more efficiently in conjunction with different types of loss functions.
The weighted combination loss function in 3.1 is expressed as a combination of Mean Square Error (MSE) and cross entropy loss:
wherein: y andreal coordinates and predicted coordinates, respectively;
s anda true category label and a predicted category label, respectively;
alpha and beta are weight parameters for balancing the two losses;
the Mean Square Error (MSE) is used to measure the squared difference between the predicted and true coordinates; the cross entropy loss is used for measuring the difference between the prediction type label and the real type label; adjusting α and β to balance the two losses;
the Adam optimizer is an optimization algorithm for deep learning application, and combines the advantages of a momentum method and RMSprop; it adjusts the learning rate based on the first moment estimate (mean) and the second moment estimate (non-centered variance) of the gradient; the update rules of Adam optimizer are:
m t ＝β 1 m t-1 +(1-β 1 )g t
/>
wherein: g t Is the gradient at time t;
m t and v t First and second moment estimates of the gradient, respectively;
β 1 and beta 2 Is an attenuation rate parameter;
θ is a model parameter;
η is the learning rate;
e is a small constant that is prevented from being divided by zero, comprising the steps of:
1. calculating the gradient: calculating a gradient of the loss function with respect to the model parameters in each iteration;
2. updating first and second moment estimates using gradient to update m t And v t ；
3. Correcting deviation: computing bias corrected moment estimatesAnd->
4. Updating parameters: and updating model parameters according to the corrected moment estimation and the self-adaptive learning rate.
In the task of automatic positioning of the heating wire, the accuracy of positioning (by MSE) and the accuracy of classification (by cross entropy) can be simultaneously optimized by using a weighted combination loss function; the Adam optimizer accelerates the convergence speed and improves the stability of the model through a self-adaptive learning rate mechanism; the training strategy enables the model to learn to extract and utilize the features from the complex images more effectively, so that the accuracy and the efficiency of heating wire positioning in an electronic cigarette automatic production line are improved.
Feedback control is a common control strategy that adjusts its input based on the output of the system to achieve the desired system performance. In automated and robotic control systems, feedback control is used to ensure accurate and stable operation.
In step 4, in the feedback control mechanism, the controller adjusts the inputs according to the output bias, assuming Δx and Δy are the heater wire position bias predicted by the deep learning model, the goal of the controller is to minimize these bias, and the control input u can be expressed as:
Wherein: e is the deviation (Δx or Δy);
K p 、K i 、K d proportional, integral, differential gains, respectively;
u is the controller output for adjusting the movement of the manipulator;
PID (proportional-integral-derivative) control is a widely used feedback control algorithm that adjusts the control input by calculating the proportional, integral and derivative of the deviation to achieve a fast and stable system response; the method comprises the following specific steps:
1. and (3) proportion control: adjusting the control input according to the current deviation;
2. and (3) integral control: accumulating past deviations for eliminating steady state errors;
3. differential control: predicting future variation of the deviation for reducing overshoot and oscillation;
in the automatic heating wire positioning system, a machine learning model and a PID control algorithm are combined to realize more accurate dynamic adjustment; the machine learning model predicts the position deviation of the heating wire, and the PID controller adjusts the movement strategy of the manipulator according to the deviation; the cooperative control strategy can improve the positioning precision and response speed, so that the overall performance and efficiency of the automatic electronic cigarette production line are improved; through closed-loop control, the system can automatically correct the deviation, ensure the accurate positioning of the heating wire, and further improve the product quality.
It should be noted that, for simplicity of description, the foregoing embodiments are all illustrated as a series of acts, but it should be understood by those skilled in the art that the present invention is not limited by the order of acts, as some steps may be performed in other order or concurrently in accordance with the present invention. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required for the present invention.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, and such partitioning of the above-described elements may be implemented in other manners, e.g., multiple elements or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or communication connection shown or discussed as being between each other may be an indirect coupling or communication connection between devices or elements via some interfaces, which may be in the form of telecommunications or otherwise.
The units described above as separate components may or may not be physically separate, and components shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the scope of the present invention. It will be apparent that the described embodiments are merely some, but not all, embodiments of the invention. Based on these embodiments, all other embodiments that may be obtained by one of ordinary skill in the art without inventive effort are within the scope of the invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art may still combine, add or delete features of the embodiments of the present invention or make other adjustments according to circumstances without any conflict, so as to obtain different technical solutions without substantially departing from the spirit of the present invention, which also falls within the scope of the present invention.
Claims (10)
1. The automatic welding device for the heating wire for producing the electronic cigarette comprises a processing table (1), a welding mechanism and a winding mechanism (2) for winding a wire on the pin end of the heating wire;
characterized in that the winding mechanism (2) comprises:
the rotary frame (21), the said rotary frame (21) is installed on processing the platform (1), there is a rotary hole (211) on the rotary frame (21) horizontally;
a rotary table (22) arranged in the rotary hole (211) in a rotary way, wherein an eccentric hole (221) is formed in the rotary table (22), and the eccentric hole (221) is not coaxial with the rotary table (22);
the sleeve (23) is arranged on the left side of the turntable (22), the sleeve (23) is fixedly connected with the turntable (22), and the sleeve (23) is coaxial with the eccentric hole (221);
the first pressing mechanism (3) is arranged on the sleeve (23) and used for pressing the right end part of the lead;
the device comprises a first platform (4) arranged on a processing table (1), wherein a semicircular groove (41) for accommodating a heating wire pin end is formed in the upper end of the first platform (4), the semicircular groove (41) is coaxial with a turntable (22), and a second pressing mechanism (5) for clamping the left end part of a wire with the heating wire pin end is arranged on the first platform (4);
and a rotation driving mechanism (6) for driving the turntable (22) to rotate.
2. The automatic welding device for heating wires for electronic cigarette production according to claim 1, wherein the first pressing mechanism (3) comprises:
The sliding sleeve (31) is elastically and slidably arranged in the sleeve (23), an arc-shaped groove (311) longitudinally extending to the inside of the sliding sleeve (31) is arranged at the top of the sliding sleeve (31), and an arc-shaped block (32) matched with the arc-shaped groove (311) is arranged in the arc-shaped groove;
the linear driving device comprises a sleeve (23) and a first linear driving source (33) arranged above the sleeve, wherein a sliding groove (231) extending along the length direction of the sliding groove is formed in the outer wall of the upper part of the sleeve (23), a sliding block (34) is connected in a sliding mode in the sliding groove (231), a strip-shaped avoidance groove (232) penetrating into the sleeve (23) is formed in the bottom of the sliding groove (231), the length direction of the strip-shaped avoidance groove (232) is consistent with the length direction of the sliding groove (231), the first linear driving source (33) is arranged on the sliding block (34), and an output shaft of the first linear driving source (33) penetrates through the sliding block (34) and penetrates through the avoidance groove to be fixedly connected with an arc-shaped block (32);
a baffle plate (35) is arranged at the left end part of the sleeve (23), and a round hole (351) which is coaxially arranged with the eccentric hole (221) is formed in the baffle plate (35);
a spring (36) is coaxially arranged in the sleeve (23), and two ends of the spring (36) are elastically abutted against the partition plate (35) and the sliding sleeve (31) respectively.
The outer wall department of sliding sleeve (31) is provided with stopper (312) along its axial extension, sleeve (23) inner wall department is provided with spacing groove (233) along its axial extension, stopper (312) sliding set up in spacing groove (233).
3. The automatic welding device for heating wires for electronic cigarette production according to claim 1, wherein the rotation driving mechanism (6) comprises:
an outer gear ring (61) fixedly sleeved on the turntable (22), wherein the outer gear ring (61) is positioned at the left side of the rotating frame (21);
a main gear (62) arranged on the left side of the rotating frame (21), wherein the main gear (62) is in meshed connection with the outer gear ring (61);
and a rotary driving source (63) fixedly installed on the right side of the turntable (22), wherein an output shaft of the rotary driving source (63) penetrates through the rotating frame (21) and is fixedly connected with the main gear (62) coaxially.
4. The automatic welding device for heating wires for electronic cigarette production according to claim 1, wherein the second pressing mechanism (5) comprises two pressing plates (51), the two pressing plates (51) are arranged in a mirror image mode on the axis of the semicircular groove (41), and the lower ends of the two pressing plates (51) are located at the same height with the top surface of the semicircular groove (41);
the outer sides of the two pressing plates (51) are respectively provided with a motor frame (52), each motor frame (52) is provided with a second linear driving source (53), and the output shaft of each second linear driving source (53) is fixedly connected to the corresponding pressing plate (51);
the opposite ends of the tops of the two pressing plates (51) are respectively provided with an inner arc-shaped plate (511), and the diameters of the two inner arc-shaped plates (511) are matched with the diameters of the wires.
5. The automatic heater strip positioning method of an automatic heater strip welding device for electronic cigarette production according to any one of claims 1 to 4, comprising the steps of:
step 1: image processing and feature extraction
1.1. Image acquisition and pretreatment:
1.11. acquiring images of the heating wire and the semicircular groove by using a high-resolution industrial camera;
1.12. segmenting the image by using a Gaussian Mixture Model (GMM), and highlighting key areas of the heating wire and the semicircular groove (41);
1.13. image enhancement using wavelet transform to improve sharpness of edge features;
1.2. feature extraction and optimization: extracting key features from the enhanced image by using a method combining Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), and constructing a feature vector quantity F for subsequent coordinate regression analysis through the features;
step 2, complex coordinate regression model
2.1. Deep learning model construction: designing a Deep Neural Network (DNN) comprising a convolution layer, a batch normalization layer, a ReLU activation layer and a full connection layer, wherein the target of the deep neural network is to predict the ideal position coordinates of the heating wire from the characteristic vector F;
2.2. coordinate regression formula: defining model outputs as delta X and delta Y, and representing deviation between an ideal position and an actual position of the heating wire;
Coordinate regression is represented using a composite function:
ΔX＝f(W 1 ·h(F)+b 1 )
ΔY＝g(W 2 ·h(F)+b 2 )
wherein f and g are nonlinear mapping functions, h is a feature extraction function, W 1 ,W 2 ,b 1 ,b 2 Is a network parameter;
step 3: model training and optimization
3.1. Loss function and optimization:
using a weighted combined loss function, combining Mean Square Error (MSE) and cross entropy loss to optimize positioning accuracy and stability; then, an Adam optimizer is applied to combine the momentum and the self-adaptive learning rate strategy so as to accelerate the convergence speed and improve the stability of the model;
step 4, system integration and cooperative control
The deviation value output by the model is used for adjusting the movement strategy of the manipulator, so that closed-loop control is realized; combines a machine learning model and a traditional PID control algorithm to realize more accurate dynamic adjustment.
6. The method for automatically positioning a heating wire of an automatic heating wire welding device for electronic cigarette production according to claim 5, wherein,
1.12. the GMM is used for image segmentation, and the heating wire and the semicircular groove are separated from the background;
the mathematical expression of the mathematical model GMM is:
wherein: p (x) is the probability of a given data point x;
k is the number of Gaussian distributions;
φ i is the mixing coefficient of the ith Gaussian distribution and meets
Is a gaussian (normal) distribution, where mu i Is the mean value, Σ i Is a covariance matrix;
first, initializing: randomly selecting parameters (mean, covariance) and mixing coefficients of K Gaussian distributions;
a desired step (E-step) of calculating the probability that each data point belongs to each Gaussian distribution;
a maximization step (M-step) of updating parameters of the Gaussian distribution and the mixing coefficients to maximize likelihood functions of the data;
and (3) carrying out substitution: repeating the E-step and the M-step until convergence (i.e., the variation of the parameters is very small or a preset number of iterations is reached);
input: image data, considered as multi-dimensional data points;
and (3) outputting: probability that each data point (pixel) belongs to each gaussian distribution is used for image segmentation;
1.13. the wavelet transformation is used for enhancing image features, particularly edges, so as to facilitate subsequent feature extraction and analysis;
the mathematical expression of the Continuous Wavelet Transform (CWT) is:
wherein x (t) is a signal;
ψ (t) is the wavelet function;
τ is a translation parameter;
s is a scale parameter;
ψ * (t) is the complex conjugate of the wavelet function;
the method comprises the following steps:
1. selecting a wavelet base: selecting a proper wavelet function;
2. calculating wavelet coefficients: for each scale and translation, computing an inner product of the signal and the wavelet basis;
3. reconstructing the signal: an optional step of reconstructing the original signal using the wavelet coefficients;
Input image data, regarded as a two-dimensional signal; wavelet coefficients are output representing features of the image at different scales and locations, which coefficients can be used for image enhancement, particularly in terms of edge detection and feature extraction.
7. The method for automatically positioning a heating wire of an automatic heating wire welding device for electronic cigarette production according to claim 5, wherein,
in 1.2, PCA is used for extracting key features from the enhanced image and reducing the dimension of the data; LDA further processes these features to maximize the degree of discrimination between the different categories (heater wire and half-slot);
mathematical model the mathematical model of the mathematical model PCA is based on a covariance matrix or Singular Value Decomposition (SVD) of the data; let X be an n X p data matrix, where n is the number of samples and p is the number of variables; PCA finds a transformation matrix W such that y=xw is a new data representation, with the columns of Y being the principal components;
the method comprises the following steps:
1. normalized data: centering and scaling the original data;
2. calculating a covariance matrix:
3. solving eigenvalues and eigenvectors: performing feature decomposition on the covariance matrix;
4. and selecting main components: selecting the first k eigenvectors according to the magnitude of the eigenvalue to form a transformation matrix W;
Inputting enhanced image data, which is regarded as multidimensional data points; outputting projection of data on a main component, namely data representation after dimension reduction;
LDA finds a linear combination W to maximize the inter-class separability while minimizing the intra-class separability; this can be achieved by maximizing the following ratio:
wherein S is B Is an inter-class scatter matrix, S W Is an intra-class scatter matrix;
the specific process is as follows:
1 calculating intra-class and inter-class scatter matrices: computing S based on class labels W And S is B ；
2 solving a generalized eigenvalue problem: finding the W that maximizes J (W);
3. projection data: projecting the data onto the found feature vector;
inputting enhanced image data, which is regarded as multidimensional data points; the projection of the output data onto the LDA feature vector, i.e. the differentiation of new feature representations of different classes.
8. The method for automatically positioning a heating wire of an automatic heating wire welding device for electronic cigarette production according to claim 5, wherein,
in 2.1, the DNN is represented as a composite of a series of functions, each corresponding to a layer in the network:
Y＝f n (...f 2 (f 1 (X,W 1 ,b 1 ),W 2 ,b 2 )...,W n ,b n )
wherein: x is input data; y is output data; f (f) i Is the activation function of the i-th layer;
W i and b i The weight and deviation of the i-th layer are respectively; the method comprises the following steps:
1. Convolution layer: extracting local features of the image using a convolution operation;
2. batch normalization layer: normalizing each small batch of inputs to improve training stability and speed;
relu activation layer: applying a nonlinear activation function ReLU (x) =max (0, x);
4. full tie layer: converting the output of the previous layer into a final output format;
inputting a feature vector F, wherein the feature vector F is obtained from the steps of image processing and feature extraction; outputting deviations delta X and delta Y of the ideal position and the actual position of the heating wire;
2.2, the purpose of the coordinate regression model is to accurately predict the deviation between the ideal position and the actual position of the heating wire; by training this model, machine learning algorithms can be made to learn to identify and correct the positional deviations of the heater filaments from the image features.
9. The method for automatically positioning a heating wire of an automatic heating wire welding device for electronic cigarette production according to claim 5, wherein,
the weighted combination loss function in 3.1 is expressed as a combination of Mean Square Error (MSE) and cross entropy loss:
wherein: y andreal coordinates and predicted coordinates, respectively;
s anda true category label and a predicted category label, respectively;
alpha and beta are weight parameters for balancing the two losses;
The Mean Square Error (MSE) is used to measure the squared difference between the predicted and true coordinates; the cross entropy loss is used for measuring the difference between the prediction type label and the real type label; adjusting α and β to balance the two losses;
the Adam optimizer is an optimization algorithm for deep learning application, and combines the advantages of a momentum method and RMSprop; it adjusts the learning rate based on the first moment estimate (mean) and the second moment estimate (non-centered variance) of the gradient; the update rules of Adam optimizer are:
m t ＝β 1 m t-1 +(1-β 1 )g t
wherein: g t Is the gradient at time t;
m t and v t First and second moment estimates of the gradient, respectively;
β 1 and beta 2 Is an attenuation rate parameter;
θ is a model parameter;
η is the learning rate;
e is a small constant that is prevented from being divided by zero, comprising the steps of:
1. calculating the gradient: calculating a gradient of the loss function with respect to the model parameters in each iteration;
2. updating first and second moment estimates using gradient to update m t And v t ；
3. Correcting deviation: computing bias corrected moment estimatesAnd->
4. Updating parameters: and updating model parameters according to the corrected moment estimation and the self-adaptive learning rate.
10. The method for automatically positioning a heating wire of an automatic heating wire welding device for electronic cigarette production according to claim 5, wherein,
In step 4, in the feedback control mechanism, the controller adjusts the inputs according to the output bias, assuming Δx and Δy are the heater wire position bias predicted by the deep learning model, the goal of the controller is to minimize these bias, and the control input u can be expressed as:
wherein: e is the deviation (Δx or Δy);
K p 、K i 、K d proportional, integral, differential gains, respectively;
u is the controller output for adjusting the movement of the manipulator;
PID (proportional-integral-derivative) control is a widely used feedback control algorithm that adjusts the control input by calculating the proportional, integral and derivative of the deviation to achieve a fast and stable system response; the method comprises the following specific steps:
1. and (3) proportion control: adjusting the control input according to the current deviation;
2. and (3) integral control: accumulating past deviations for eliminating steady state errors;
3. differential control: predicting future variation of the deviation for reducing overshoot and oscillation;
in the automatic heating wire positioning system, a machine learning model and a PID control algorithm are combined to realize more accurate dynamic adjustment; the machine learning model predicts the position deviation of the heating wire, and the PID controller adjusts the movement strategy of the manipulator according to the deviation; the cooperative control strategy can improve the positioning precision and response speed, so that the overall performance and efficiency of the automatic electronic cigarette production line are improved; through closed-loop control, the system can automatically correct the deviation, ensure the accurate positioning of the heating wire, and further improve the product quality.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311582552.3A CN117483996B (en) | 2023-11-24 | 2023-11-24 | Automatic welding device and automatic positioning method for heating wire for electronic cigarette production |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311582552.3A CN117483996B (en) | 2023-11-24 | 2023-11-24 | Automatic welding device and automatic positioning method for heating wire for electronic cigarette production |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117483996A true CN117483996A (en) | 2024-02-02 |
| CN117483996B CN117483996B (en) | 2024-07-02 |
Family
ID=89668963
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311582552.3A Active CN117483996B (en) | 2023-11-24 | 2023-11-24 | Automatic welding device and automatic positioning method for heating wire for electronic cigarette production |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117483996B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6957615B1 (en) * | 1999-05-14 | 2005-10-25 | Meca S.P.A. | Method and device to apply cord thread or ribbons onto fabrics in a quilting machine |
| CN207074582U (en) * | 2017-08-03 | 2018-03-06 | 常州贝伦迪线束技术股份有限公司 | A kind of wire harness wrapping machine |
| CN208954798U (en) * | 2018-12-19 | 2019-06-07 | 常熟市吉宏汽车电子科技有限公司 | A kind of electromagnetic coil lead-out wire winding tool |
| CN209815361U (en) * | 2019-03-22 | 2019-12-20 | 定州市五星金属网厂 | Galvanized wire rolling machine |
| CN112758769A (en) * | 2021-01-13 | 2021-05-07 | 河南知果信息科技有限公司 | A weaving yarn kinking winding mechanism for textile processing |
| CN215658571U (en) * | 2021-06-21 | 2022-01-28 | 深圳市锐博精创科技有限公司 | Welding equipment for heating wire |
| CN116338571A (en) * | 2023-03-06 | 2023-06-27 | 中山大学 | RSSI fingerprint positioning method based on self-encoder and attention mechanism |
| CN116441783A (en) * | 2023-06-06 | 2023-07-18 | 深圳市大道四九科技有限公司 | Automatic welding device for heating wire for electronic cigarette production |
| WO2023159298A1 (en) * | 2022-02-28 | 2023-08-31 | National Research Council Of Canada | Deep learning based prediction of fabrication-process-induced structural variations in nanophotonic devices |
| CN116958584A (en) * | 2023-09-21 | 2023-10-27 | 腾讯科技（深圳）有限公司 | Key point detection method, regression model training method and device and electronic equipment |
-
2023
- 2023-11-24 CN CN202311582552.3A patent/CN117483996B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6957615B1 (en) * | 1999-05-14 | 2005-10-25 | Meca S.P.A. | Method and device to apply cord thread or ribbons onto fabrics in a quilting machine |
| CN207074582U (en) * | 2017-08-03 | 2018-03-06 | 常州贝伦迪线束技术股份有限公司 | A kind of wire harness wrapping machine |
| CN208954798U (en) * | 2018-12-19 | 2019-06-07 | 常熟市吉宏汽车电子科技有限公司 | A kind of electromagnetic coil lead-out wire winding tool |
| CN209815361U (en) * | 2019-03-22 | 2019-12-20 | 定州市五星金属网厂 | Galvanized wire rolling machine |
| CN112758769A (en) * | 2021-01-13 | 2021-05-07 | 河南知果信息科技有限公司 | A weaving yarn kinking winding mechanism for textile processing |
| CN215658571U (en) * | 2021-06-21 | 2022-01-28 | 深圳市锐博精创科技有限公司 | Welding equipment for heating wire |
| WO2023159298A1 (en) * | 2022-02-28 | 2023-08-31 | National Research Council Of Canada | Deep learning based prediction of fabrication-process-induced structural variations in nanophotonic devices |
| CN116338571A (en) * | 2023-03-06 | 2023-06-27 | 中山大学 | RSSI fingerprint positioning method based on self-encoder and attention mechanism |
| CN116441783A (en) * | 2023-06-06 | 2023-07-18 | 深圳市大道四九科技有限公司 | Automatic welding device for heating wire for electronic cigarette production |
| CN116958584A (en) * | 2023-09-21 | 2023-10-27 | 腾讯科技（深圳）有限公司 | Key point detection method, regression model training method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117483996B (en) | 2024-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106778863A (en) | The warehouse kinds of goods recognition methods of dictionary learning is differentiated based on Fisher | |
| CN110303380A (en) | A kind of cutting tool for CNC machine method for predicting residual useful life | |
| CN109670580A (en) | A kind of data recovery method based on time series | |
| EP4107588A1 (en) | Systems, methods, and media for manufacturing processes | |
| CN117483996B (en) | Automatic welding device and automatic positioning method for heating wire for electronic cigarette production | |
| CN111310791A (en) | Dynamic progressive automatic target identification method based on small sample number set | |
| US20210311440A1 (en) | Systems, Methods, and Media for Manufacturing Processes | |
| CN112966735B (en) | Method for fusing supervision multi-set related features based on spectrum reconstruction | |
| CN113469262A (en) | Incremental learning method based on Fisher information matrix | |
| CN109190505A (en) | The image-recognizing method that view-based access control model understands | |
| CN113837046A (en) | Small sample remote sensing image scene classification method based on iterative feature distribution learning | |
| CN117993282A (en) | Domain adaptability information bottleneck federal learning method for intelligent manufacturing fault diagnosis | |
| CN112686268A (en) | Crop leaf disorder identification method based on SVD-ResNet50 neural network | |
| CN109522914A (en) | A kind of neural network structure training method of the Model Fusion based on image | |
| CN116755811A (en) | Desktop virtualization method and system for fusion application | |
| CN109165587B (en) | Intelligent image information extraction method | |
| CN116665039A (en) | Small sample target identification method based on two-stage causal intervention | |
| Mototake et al. | Spectral deconvolution through Bayesian LARS-OLS | |
| Ündey et al. | Monitoring of batch pharmaceutical fermentations: Data synchronization, landmark alignment, and real-time monitoring | |
| CN114943862A (en) | Two-stage image classification method based on structural analysis dictionary learning | |
| CN109344881B (en) | Extended classifier based on space-time continuity | |
| CN104636758B (en) | A kind of SAR image suitability Forecasting Methodology based on support vector regression | |
| CN114037860A (en) | Image classification and feature selection method based on robust minimum two-regression framework | |
| CN111160464B (en) | Industrial high-order dynamic process soft measurement method based on multi-hidden-layer weighted dynamic model | |
| CN109934304B (en) | Blind domain image sample classification method based on out-of-limit hidden feature model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |