Optimization method and system of hot melt self-tapping connection based on artificial neural network

Abstract

This invention discloses an optimization method and system for hot-melt self-tapping screw connections based on artificial neural networks, comprising: (1) obtaining an image database of the hot-melt self-tapping screw connection joint cross-section through non-destructive testing; (2) performing visual recognition on the images of the hot-melt self-tapping screw connection joint cross-section; (3) establishing a database of multiple sets of parameters for inter-plate clearance and effective thread engagement length, and constructing a nonlinear mapping relationship prediction model between hot-melt self-tapping screw connection process parameters and joint quality through artificial neural networks; (4) obtaining process parameters at different stages during the connection process between the hot-melt self-tapping screw and the material through a monitoring module; and (5) sending the correct process parameters to the hot-melt self-tapping screw automatic tightening device to control the operation of the hot-melt self-tapping screw automatic tightening device in real time. The hot-melt self-tapping screw connection optimization method and system based on artificial neural networks of this invention can optimize the hot-melt self-tapping screw connection process and improve the quality of the hot-melt self-tapping screw connection joint forming.

Description

Technical Field

[0001] The invention relates to an optimization method and an optimization system for hot-melt self-tapping connection based on an artificial neural network, belonging to the technical field of hot-melt self-tapping connection molding for connecting dissimilar materials. Background Art

[0002] With the development of the automobile industry, new materials with high mechanical properties and low density are being used more and more widely. In order to achieve lightweight automobile bodies, the application of multiple materials is becoming a trend in the manufacturing industry, and new structures are constantly emerging, such as integrated die-cast body parts. It is precisely because of the emergence of various new materials and new structures that the original electric welding technology suitable for all-steel bodies is no longer applicable. In order to achieve the connection of new materials, riveting and screwing have become new development directions. Facing the connection between die-cast aluminum and extruded aluminum, high-strength steel and extruded aluminum, hot-melt self-tapping screws (FDS) have demonstrated their unique superiority, especially in the connection between dissimilar materials in closed cavities. The most typical example is the connection between integrated die-cast aluminum body parts and extruded aluminum, and the connection between high-strength steel B-pillars and extruded aluminum is generally connected by hot-melt self-tapping screws.

[0003] However, how to achieve hot-melt self-tapping (FDS) connections with different material properties and thicknesses requires adjusting different connection parameters. How to reduce the time and workload of establishing the optimal parameters is a problem that needs to be considered.

[0004] In the process of hot melt self-tapping connection, FDS forming includes six main stages:

[0005] 1. Positioning stage: The flow drill screw contacts the plate surface at low speed and low pressure;

[0006] 2. Penetration stage: The plate undergoes plastic deformation under the high speed and high pressure of the flow drill screw;

[0007] 3. Hole forming stage: The flow drill screw drills a cylindrical through hole in the plate;

[0008] 4. Thread forming stage: the plate is drilled by the flow drill screw, the speed and pressure are reduced, and the flow drill screw self-taps the thread in the hole;

[0009] 5. Thread tightening stage: normal screw tightening process;

[0010] 6. Fastening stage: The panels cool down and fit tightly with the flow drill screws, forming a highly airtight and watertight connection;

[0011] In the process of hot melt self-tapping connection, the transition from one step to another should be precisely controlled and carried out purposefully. For example, the transition from penetration forming to hole forming should be accurately detected and controlled. At this stage, the process parameters (especially the number of revolutions, but also process variables such as clamping force and / or feed rate) have a key influence on the quality of hole and thread forming. Tapping is only started when the drilling depth is large enough to ensure the quality of hole and thread forming. Starting the screw rotation too early or too late will affect the thread formation. The transition between different stages does not have to correspond to a specific moment, but can be transitioned to tapping in a continuous process by adjusting or controlling the characteristics. Similar situations also apply to the transition between other steps of the hot melt self-tapping process.

[0012] In the prior art, the switching moments of the different stages are usually determined by measuring the drilling depth. However, in this assumption, the thickness error of the workpiece is not taken into account. If the screw penetrates the workpiece too early, the tapping part of the screw may be inserted into the workpiece with too high a number of revolutions and too large a feed force, thus damaging the thread.

[0013] Generally speaking, the mechanical properties of FDS joints are the key to evaluating their joint quality. However, the existing technology only uses a universal testing machine to perform multiple sets of repeated stretching on a single joint to measure the maximum load that the joint can bear. This is obviously not enough for multiple sets of joints with different material combinations and large-scale component-level FDS joint clusters, and the excessive workload makes it difficult to achieve the highest verification efficiency in engineering applications. Given that the degree of meshing of FDS joints is linearly related to the mechanical properties, but considering that the orthogonal test requires cutting each section to determine the degree of meshing of the FDS joint, the workload of establishing the optimal process parameters is further increased, and the efficiency of establishing the optimal process parameters is not high.

[0014] The patent with the authorization announcement number CN 211235243 U discloses an oblique tensile test fixture for a hot-melt self-tapping connection joint, comprising: a first sample fixing block, which has a first sample connection end; a first fixing rod, one end of which is connected to the first sample fixing block, and the other end of the first fixing rod can be connected to the stretch sleeve of the tensile testing machine, and the first sample connection end and the first fixing rod are at an oblique angle; a second sample fixing block, which has a second sample connection end; a second fixing rod, one end of which is connected to the second sample fixing block, and the other end of the second fixing rod can be connected to another stretch sleeve of the tensile testing machine, and the second sample connection end and the second fixing rod are at an oblique angle. The oblique tensile test fixture can make the hot-melt self-tapping connection joint be fixed at a certain angle between the two stretch sleeves of the tensile testing machine, so as to facilitate the use of an ordinary tensile testing machine to perform oblique tensile mechanical property tests on the hot-melt self-tapping connection joint. However, the patent does not solve the problem of optimizing the hot-melt self-tapping connection process. Summary of the invention

[0015] The purpose of the present invention is to provide a hot-melt self-tapping connection optimization method and optimization system based on artificial neural network. The hot-melt self-tapping connection optimization method and optimization system based on artificial neural network can optimize the hot-melt self-tapping connection process and improve the quality of hot-melt self-tapping connection joint forming.

[0016] In order to solve the above technical problems, a technical solution adopted by the present invention is: to provide a hot melt self-tapping connection optimization method based on artificial neural network, comprising the following steps:

[0017] (1) Obtaining an image database of the cross-section of hot-melt self-tapping connection joints through non-destructive testing;

[0018] (2) Visually identify the cross-section image of the hot-melt self-tapping connection joint, and output the inter-plate gap and effective thread engagement length parameters between the upper plate and the lower plate when the hot-melt self-tapping screw is connected to the upper plate and the lower plate;

[0019] (3) Establish a database of multiple sets of inter-plate gaps and effective thread engagement length parameters, and use artificial neural networks to construct a nonlinear mapping relationship prediction model between hot-melt self-tapping connection process parameters and joint quality, and determine the quality acceptance standard of hot-melt self-tapping connection joints;

[0020] (4) obtaining the process parameters of the hot-melt self-tapping screws at different stages of the process of connecting the hot-melt self-tapping screws to the materials through the monitoring module, inputting these process parameters into the prediction model to determine whether the quality of the hot-melt self-tapping screw connection joint is qualified, and giving the correct process parameters to the control module through the prediction model;

[0021] (5) The control module sends the correct process parameters to the hot melt self-tapping automatic tightening device to control the operation of the hot melt self-tapping automatic tightening device in real time.

[0022] Preferably, step (1) includes: setting up a nondestructive testing module, which uses CT / X-rays to perform nondestructive testing on the connection joints between the hot-melt self-tapping screws and the upper and lower plates, obtains a cross-sectional image of the connection joints between the hot-melt self-tapping screws and the upper and lower plates, and stores it as an initial image.

[0023] Preferably, step (2) includes: setting a visual recognition module, the visual recognition module extracts an initial image, performs visual recognition on the initial image, and outputs the inter-plate gap and effective thread engagement length parameters of the upper plate and the lower plate, wherein when the effective thread engagement length is defined, the screw thread and the stud tooth surface are less than or equal to 5 microns, which is an effective thread, otherwise it is an invalid thread, the inter-plate gap between the upper plate and the lower plate is stored as d1, the effective thread engagement length is stored as d2, and the joint forming quality of the hot-melt self-tapping screw connecting the upper plate and the lower plate is linearly related to d1 and d2, the smaller d1 and the larger d2, the higher the joint forming quality, so when there is an inter-plate gap on both sides, d1 takes the end with a larger absolute value, and the effective thread engagement length on the left and right sides takes the end with a smaller absolute value, and the presence or absence of cracks is stored as A, 0 if there is a crack, and 1 if there is no crack.

[0024] Preferably, step (3) includes: repeating the operations of step (1) and step (2) to obtain cross-sectional images of hot-melt self-tapping connection joints when multiple groups of different process parameters are connected, and extracting multiple groups of d1, d2 and A data, establishing a database, setting a machine learning module, and constructing a nonlinear mapping relationship prediction model between hot-melt self-tapping connection process parameters and joint quality through an artificial neural network. The prediction model formula is: the model input of the artificial neural network includes the down force n1, the rotation speed n2, the tightening force n3, and the experience time of the three stages of penetration time t1, hole forming time t2 and thread forming time t3. The three stages are seamlessly connected, and the transition time is set to 0. The model output is d1*, d2*, A*. The output d1* and d2* are the predicted inter-plate gap and effective meshing thread length. A is "whether there is a crack". If there is a crack, it is 0, and if there is no crack, it is 1, wherein:

[0025] d1*=f1(n1,n2,n3,t1,t2,t3);

[0026] d2*=f2(n1,n2,n3,t1,t2,t3);

[0027] A*=f3(n1,n2,n3,t1,t2,t3);

[0028] The quality evaluation factor Q is linearly related to d1* and d2*. The smaller d1* is, the larger d2* is. The larger the Q value is, the higher the joint forming quality is. However, cracks are not allowed to exist for the joint quality. Therefore, as long as there are cracks, the joint is unqualified. Therefore, the relationship between Q and d1*, d2*, and A* is:

[0029] Q=A*×(k1×d1*+k2×d2*)+C, where k1, k2, and C are constant values.

[0030] Preferably, according to the hot melt self-tapping connection optimization method based on artificial neural network according to claim 4, it is characterized in that step (4) includes: connecting the prediction module and the comparison module of the prediction model to each other, and calculating the dynamic gain σ1 of the down force and the dynamic gain σ2 of the speed, the dynamic gain σ3 of the tightening force, the penetration time gain σ4, the hole forming time gain σ5, and the thread forming time gain σ6 at the current moment according to the current die-casting quality Q1 connected to the hot melt self-tapping and the optimal die-casting quality Q0, and obtaining the optimized injection force λ1, the injection speed λ2 at the current moment, the current moment λ3, the penetration time λ4 at the current moment, the hole forming time λ5 at the current moment, and the thread forming time λ6 at the current moment, then:

[0031] λm=σ1·nm / tm(m=1,2,3);

[0032] Finally, after the threshold judgment in the comparison module, the quality factor obtained by the initially input process parameters is compared with the database, and the quality factor 0.8·Q0 is defined as the qualified limit value. If it is greater than 0.8·Q0, the output is qualified, and if it is less than 0.8·Q0, the output is unqualified. While outputting the qualified signal, the hot-melt self-tapping connection process parameters are automatically optimized, and finally the optimized process parameter combination is obtained.

[0033] Preferably, step (5) includes: the control module extracts the parameters of the down force n1, rotation speed n2, tightening force n3, penetration time t1, hole forming time t2 and thread forming time t3 of the comparison module and the prediction module, and sends the above parameter signal values ​​to the hot-melt self-tapping automatic tightening device for real-time control of the hot-melt self-tapping connection action process.

[0034] The present invention also provides an optimization system for a hot-melt self-tapping connection optimization method based on an artificial neural network, comprising: a monitoring module for monitoring process parameters at different stages during the connection between the hot-melt self-tapping thread and the material; a visual recognition module for visually identifying images of the cross-section of the hot-melt self-tapping connection joint; a prediction module and a comparison module for predicting and comparing the quality of the hot-melt self-tapping connection joint; and a control module for extracting the optimized hot-melt self-tapping connection process parameters in the comparison module and the prediction module, and sending the optimized hot-melt self-tapping connection process parameters to the hot-melt self-tapping automatic tightening device for execution.

[0035] Preferably, it also includes non-destructive testing equipment, the monitoring module and the control module are arranged on the hot-melt self-tapping automatic tightening device, the visual recognition module is arranged on the non-destructive testing equipment, and the prediction module and the comparison module are arranged on the hot-melt self-tapping automatic tightening device.

[0036] The beneficial effects of the present invention are as follows: the hot-melt self-tapping connection optimization method and optimization system based on artificial neural network of the present invention construct a nonlinear mapping relationship prediction model between the hot-melt self-tapping connection process parameters and the joint quality through artificial neural network, monitor the process parameters at different stages in the process of connecting the hot-melt self-tapping thread with the material through the monitoring module, and perform visual recognition on the picture of the cross-section of the hot-melt self-tapping connection joint through the visual recognition module; predict and compare the quality of the hot-melt self-tapping connection joint through the prediction module and the comparison module; extract the optimized hot-melt self-tapping connection process parameters in the comparison module and the prediction module through the control module, and send the optimized hot-melt self-tapping connection process parameters to the hot-melt self-tapping automatic tightening device for execution, thereby realizing the optimization of the hot-melt self-tapping connection process and improving the quality of the hot-melt self-tapping connection joint forming. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0038] Figure 1 Schematic diagram of the steps of the hot melt self-tapping connection optimization method based on artificial neural network of the present invention;

[0039] Figure 2 It is a schematic diagram of an optimization system of the hot-melt self-tapping connection optimization method based on artificial neural network of the present invention. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] Embodiments of the present invention:

[0042] The present invention provides a method for optimizing hot melt self-tapping connection based on artificial neural network. Figure 1 , including the following steps:

[0043] (1) Obtaining an image database of the cross-section of hot-melt self-tapping connection joints through non-destructive testing;

[0044] (2) Visually identify the cross-section image of the hot-melt self-tapping connection joint, and output the inter-plate gap and effective thread engagement length parameters between the upper plate and the lower plate when the hot-melt self-tapping screw is connected to the upper plate and the lower plate;

[0045] (3) Establish a database of multiple sets of inter-plate gaps and effective thread engagement length parameters, and use artificial neural networks to construct a nonlinear mapping relationship prediction model between hot-melt self-tapping connection process parameters and joint quality, and determine the quality acceptance standard of hot-melt self-tapping connection joints;

[0046] (4) obtaining the process parameters of the hot-melt self-tapping screws at different stages of the process of connecting the hot-melt self-tapping screws to the materials through the monitoring module, inputting these process parameters into the prediction model to determine whether the quality of the hot-melt self-tapping screw connection joint is qualified, and giving the correct process parameters to the control module through the prediction model;

[0047] (5) The control module sends the correct process parameters to the hot melt self-tapping automatic tightening device to control the operation of the hot melt self-tapping automatic tightening device in real time.

[0048] Preferably, step (1) includes: setting up a nondestructive testing module, which uses CT / X-rays to perform nondestructive testing on the connection joints between the hot-melt self-tapping screws and the upper and lower plates, obtains a cross-sectional image of the connection joints between the hot-melt self-tapping screws and the upper and lower plates, and stores it as an initial image.

[0049] Preferably, step (2) includes: setting a visual recognition module, the visual recognition module extracts an initial image, performs visual recognition on the initial image, and outputs the inter-plate gap and effective thread engagement length parameters of the upper plate and the lower plate, wherein when the effective thread engagement length is defined, the screw thread and the stud tooth surface are less than or equal to 5 microns, which is an effective thread, otherwise it is an invalid thread, the inter-plate gap between the upper plate and the lower plate is stored as d1, the effective thread engagement length is stored as d2, and the joint forming quality of the hot-melt self-tapping screw connecting the upper plate and the lower plate is linearly related to d1 and d2, the smaller d1 and the larger d2, the higher the joint forming quality, so when there is an inter-plate gap on both sides, d1 takes the end with a larger absolute value, and the effective thread engagement length on the left and right sides takes the end with a smaller absolute value, and the presence or absence of cracks is stored as A, 0 if there is a crack, and 1 if there is no crack.

[0050] In the process of hot-melt self-tapping drilling into materials, the key process parameters include downward pressure, rotation speed, tightening pitch, and the start and transition time of the penetration, hole forming and thread forming stages. The non-destructive testing module obtains the cross-sectional view of the FDS meshing through CT, X-ray and ultrasonic non-destructive testing technology scanning, and obtains the gap between plates, effective thread engagement length, and defects such as cracks and pits through the visual recognition module.

[0051] Preferably, step (3) includes: repeating the operations of step (1) and step (2) to obtain cross-sectional images of hot-melt self-tapping connection joints when multiple groups of different process parameters are connected, and extracting multiple groups of d1, d2 and A data, establishing a database, setting a machine learning module, and constructing a nonlinear mapping relationship prediction model between hot-melt self-tapping connection process parameters and joint quality through an artificial neural network. The prediction model formula is: the model input of the artificial neural network includes the down force n1, the rotation speed n2, the tightening force n3, and the experience time of the three stages of penetration time t1, hole forming time t2 and thread forming time t3. The three stages are seamlessly connected, and the transition time is set to 0. The model output is d1*, d2*, A*. The output d1* and d2* are the predicted inter-plate gap and effective meshing thread length. A is "whether there is a crack". If there is a crack, it is 0, and if there is no crack, it is 1, wherein:

[0052] d1*=f1(n1,n2,n3,t1,t2,t3);

[0053] d2*=f2(n1,n2,n3,t1,t2,t3);

[0054] A*=f3(n1,n2,n3,t1,t2,t3);

[0055] The quality evaluation factor Q is linearly related to d1* and d2*. The smaller d1* is, the larger d2* is. The larger the Q value is, the higher the joint forming quality is. However, cracks are not allowed to exist for the joint quality. Therefore, as long as there are cracks, the joint is unqualified. Therefore, the relationship between Q and d1*, d2*, and A* is:

[0056] Q=A*×(k1×d1*+k2×d2*)+C, where k1, k2, and C are constant values.

[0057] Preferably, according to the hot melt self-tapping connection optimization method based on artificial neural network according to claim 4, it is characterized in that step (4) includes: connecting the prediction module and the comparison module of the prediction model to each other, and calculating the dynamic gain σ1 of the down force and the dynamic gain σ2 of the speed, the dynamic gain σ3 of the tightening force, the penetration time gain σ4, the hole forming time gain σ5, and the thread forming time gain σ6 at the current moment according to the current die-casting quality Q1 connected to the hot melt self-tapping and the optimal die-casting quality Q0, and obtaining the optimized injection force λ1, the injection speed λ2 at the current moment, the current moment λ3, the penetration time λ4 at the current moment, the hole forming time λ5 at the current moment, and the thread forming time λ6 at the current moment, then:

[0058] λm=σ1·nm / tm(m=1,2,3);

[0059] Finally, after the threshold judgment in the comparison module, the quality factor obtained by the initially input process parameters is compared with the database, and the quality factor 0.8·Q0 is defined as the qualified limit value. If it is greater than 0.8·Q0, the output is qualified, and if it is less than 0.8·Q0, the output is unqualified. While outputting the qualified signal, the hot-melt self-tapping connection process parameters are automatically optimized, and finally the optimized process parameter combination is obtained.

[0060] Preferably, step (5) includes: the control module extracts the parameters of the down force n1, rotation speed n2, tightening force n3, penetration time t1, hole forming time t2 and thread forming time t3 of the comparison module and the prediction module, and sends the above parameter signal values ​​to the hot-melt self-tapping automatic tightening device for real-time control of the hot-melt self-tapping connection action process.

[0061] A database is established for the key data of the cross-section of hot-melt self-tapping joints formed by different process parameters. Through machine learning, the key process parameters and joint quality assessment parameters are finally fitted into a nonlinear mapping relationship. The nonlinear mapping relationship of the quality prediction model of hot-melt self-tapping joints can be constructed by artificial neural networks or numerical fitting prediction formulas. The prediction results include the gap between plates, the effective thread engagement length, "whether there is a crack", and online judgment of whether it is qualified.

[0062] In order to optimize the process parameters, the process parameters can be introduced into the artificial neural network. The monitoring module records the downforce, speed, tightening force and time of each stage, and forms an image with time as the horizontal axis and downforce, speed and tightening force as the vertical axis, so as to better distinguish the changes in process parameters in each stage and better realize the transition between stages.

[0063] For a specific material combination, the initial six process parameters of the hot-melt self-tapping joint of the specific material combination are obtained by monitoring, and the two key parameters for evaluating the joint quality, namely the inter-plate gap and the effective meshing thread length, as well as the presence of cracks, which is an unacceptable defect, can be output. Finally, by comparing the FDS joint quality factors of different process parameters actually produced in the database, it is directly possible to determine whether the process parameter combination is qualified. Further artificial neural network prediction and iteration are used to obtain a better process parameter combination, and the final output is a process parameter combination with optimized molding quality. The control module extracts the down force n1, speed n2, tightening force n3 of the comparison module and the prediction module, as well as the experience time of the three stages of penetration t1, hole forming t2 and thread forming t3, and sends the above signal values ​​to the hot-melt self-tapping automatic tightening device for real-time control of the material connection process.

[0064] The present invention also provides an optimization system for a hot-melt self-tapping connection optimization method based on an artificial neural network, comprising: a monitoring module for monitoring process parameters at different stages during the connection between the hot-melt self-tapping thread and the material; a visual recognition module for visually identifying images of the cross-section of the hot-melt self-tapping connection joint; a prediction module and a comparison module for predicting and comparing the quality of the hot-melt self-tapping connection joint; and a control module for extracting the optimized hot-melt self-tapping connection process parameters in the comparison module and the prediction module, and sending the optimized hot-melt self-tapping connection process parameters to the hot-melt self-tapping automatic tightening device for execution.

[0065] Preferably, it also includes non-destructive testing equipment, the monitoring module and the control module are arranged on the hot-melt self-tapping automatic tightening device, the visual recognition module is arranged on the non-destructive testing equipment, and the prediction module and the comparison module are arranged on the hot-melt self-tapping automatic tightening device.

[0066] The optimization system of the present invention executes the hot-melt self-tapping connection optimization method based on artificial neural network to achieve the optimization processing of the hot-melt self-tapping connection.

[0067] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for optimizing hot-melt self-tapping connections based on artificial neural networks, characterized in that: The following steps are involved: (1) Obtaining an image database of the cross-section of hot-melt self-tapping connection joints through non-destructive testing; (2) Visually identify the cross-section image of the hot-melt self-tapping connection joint, and output the inter-plate gap and effective thread engagement length parameters between the upper plate and the lower plate when the hot-melt self-tapping screw is connected to the upper plate and the lower plate; (3) Establish a database of multiple sets of inter-plate gaps and effective thread engagement length parameters, and use artificial neural networks to construct a nonlinear mapping relationship prediction model between hot-melt self-tapping connection process parameters and joint quality, and determine the quality acceptance standard of hot-melt self-tapping connection joints; (4) obtaining the process parameters of the hot-melt self-tapping screws at different stages of the process of connecting the hot-melt self-tapping screws to the materials through the monitoring module, inputting these process parameters into the prediction model to determine whether the quality of the hot-melt self-tapping screw connection joint is qualified, and giving the correct process parameters to the control module through the prediction model; (5) The control module sends the correct process parameters to the hot melt self-tapping automatic tightening device to control the operation of the hot melt self-tapping automatic tightening device in real time.

2. The method for optimizing hot-melt self-tapping connection based on artificial neural network according to claim 1 is characterized in that: Step (1) includes: setting up a nondestructive testing module, which uses CT / X-rays to perform nondestructive testing on the connection joints between the hot-melt self-tapping screws and the upper and lower plates, obtains a cross-sectional image of the connection joints between the hot-melt self-tapping screws and the upper and lower plates, and stores it as an initial image.

3. The method for optimizing hot-melt self-tapping connection based on artificial neural network according to claim 2 is characterized in that: Step (2) includes: setting a visual recognition module, the visual recognition module extracts an initial image, performs visual recognition on the initial image, and outputs the inter-plate gap and effective thread engagement length parameters of the upper plate and the lower plate, wherein when the effective thread engagement length is defined, the screw thread and the stud tooth surface are less than or equal to 5 microns, which is an effective thread, otherwise it is an invalid thread, the inter-plate gap between the upper plate and the lower plate is stored as d1, the effective thread engagement length is stored as d2, and the joint forming quality of the hot-melt self-tapping screw connecting the upper plate and the lower plate is linearly related to d1 and d2, the smaller d1 and the larger d2, the higher the joint forming quality, so when there is an inter-plate gap on both sides, d1 takes the end with a larger absolute value, and the effective thread engagement length on the left and right sides takes the end with a smaller absolute value, and the presence or absence of cracks is stored as A, where the presence of cracks is 0 and the absence of cracks is 1.

4. The method for optimizing hot-melt self-tapping connection based on artificial neural network according to claim 3 is characterized in that: Step (3) includes: repeating the operations of step (1) and step (2) to obtain cross-sectional images of hot-melt self-tapping connection joints when multiple groups of different process parameters are connected, and extracting multiple groups of d1, d2 and A data, establishing a database, setting a machine learning module, and constructing a nonlinear mapping relationship prediction model between hot-melt self-tapping connection process parameters and joint quality through an artificial neural network. The prediction model formula is: the model input of the artificial neural network includes the down force n1, the rotation speed n2, the tightening force n3, and the experience time of the three stages of penetration time t1, hole forming time t2 and thread forming time t3. The three stages are seamlessly connected, and the transition time is set to 0. The model output is d1*, d2*, A*. The output d1* and d2* are the predicted inter-plate gap and effective meshing thread length. A is "whether there is a crack". If there is a crack, it is 0, and if there is no crack, it is 1. Among them: d1*=f1(n1,n2,n3,t1,t2,t3); d2*=f2(n1, n2, n3, t1, t2, t3); A*=f3(n1,n2,n3,t1,t2,t3); f1, f2, f3 represent three different function mapping relationships, corresponding to the mapping relationships between d1*, d2*, A* and n1, n2, n3, t1, t2, t3 respectively; The quality evaluation factor Q is linearly related to d1* and d2*. The smaller d1* is, the larger d2* is. The larger the Q value is, the higher the joint forming quality is. However, cracks are not allowed to exist for the joint quality. Therefore, as long as there are cracks, the joint is unqualified. Therefore, the relationship between Q and d1*, d2*, and A* is: Q=A*×(k1×d1*+k2×d2*)+C, where k1, k2, and C are constant values.

5. The method for optimizing hot-melt self-tapping connection based on artificial neural network according to claim 4 is characterized in that: Step (4) includes: connecting the prediction module and the comparison module of the prediction model to each other, calculating the down force dynamic gain σ1 and the speed dynamic gain σ2, the tightening force dynamic gain σ3, the penetration time gain σ4, the hole forming time gain σ5, and the thread forming time gain σ6 at the current moment according to the die casting quality Q1 of the current hot melt self-tapping connection and the optimal die casting quality Q0, and obtaining the optimized injection force λ1, the injection speed λ2 at the current moment, the current moment λ3, the penetration time λ4 at the current moment, the hole forming time λ5 at the current moment, and the thread forming time λ6 at the current moment, then: Xm is the adjustment coefficient of the six parameters; Finally, after the threshold judgment in the comparison module, the quality evaluation factor obtained by the initially input process parameters is compared with the database, and the quality evaluation factor 0.8·Q0 is defined as the qualified limit value. If it is greater than 0.8·Q0, the output is qualified, and if it is less than 0.8·Q0, the output is unqualified. While outputting the qualified or not signal, the hot-melt self-tapping connection process parameters are automatically optimized, and finally the optimized process parameter combination is obtained.

6. The method for optimizing hot-melt self-tapping connection based on artificial neural network according to claim 5 is characterized in that: Step (5) includes: the control module extracts the parameters of the down force n1, rotation speed n2, tightening force n3, penetration time t1, hole forming time t2 and thread forming time t3 from the comparison module and the prediction module, and sends the above parameter signal values ​​to the hot melt self-tapping automatic tightening device for real-time control of the hot melt self-tapping connection action process.

7. An optimization system for the hot melt self-tapping connection optimization method based on artificial neural network according to any one of claims 1 to 6, characterized in that: include: Monitoring module for monitoring process parameters at different stages of the process of hot-melt self-tapping screws connecting with materials; A visual recognition module for visually identifying a picture of a cross section of a hot-melt self-tapping connection joint; a prediction module and a comparison module for predicting and comparing the quality of a hot-melt self-tapping connection joint; A control module used to extract the optimized hot-melt self-tapping connection process parameters in the comparison module and the prediction module, and send the optimized hot-melt self-tapping connection process parameters to the hot-melt self-tapping automatic tightening device for execution.

8. The optimization system according to claim 7, characterized in that: It also includes non-destructive testing equipment, a monitoring module and a control module are arranged on the hot-melt self-tapping automatic tightening device, a visual recognition module is arranged on the non-destructive testing equipment, and a prediction module and a comparison module are arranged on the hot-melt self-tapping automatic tightening device.

Images