Ultrasonic welding quality monitoring system and method using artificial intelligence

Abstract

The present technology relates to an ultrasonic welding quality monitoring system and method using artificial intelligence, which collects the pattern of voltage signals generated during ultrasonic welding and can effectively inspect the welding quality therefrom.

Description

Technology Field

[0001] This invention relates to an ultrasonic welding quality monitoring system and method using artificial intelligence. Background Technology

[0003] As technology for mobile devices advances, the demand for rechargeable secondary batteries as power sources is rapidly increasing, and much research is being conducted to meet the diverse technological development requirements for these batteries.

[0004] In terms of form factor, there is high demand for pouch-type and prismatic secondary batteries, which can be applied to products such as mobile phones due to their thin thickness.

[0005] There is high demand for lithium secondary batteries, such as lithium-ion batteries and lithium-ion polymer batteries, which have advantages in terms of materials, such as high energy density, discharge voltage, and output stability. Lithium secondary batteries have an operating voltage of about 3.6V, which is about three times the capacity of nickel-cadmium or nickel-hydrogen batteries, and their utilization is rapidly increasing because of their high energy density per unit weight.

[0006] Secondary batteries are sometimes classified according to the structure of the electrode assembly. Examples include jelly-roll wound electrode assemblies, which consist of long sheet-type positive and negative electrodes wound with a separator; stacked electrode assemblies, which consist of multiple positive and negative electrodes cut into units of a predetermined size and sequentially stacked with a separator; and stack-and-folding electrode assemblies, which consist of bi-cells or full cells formed by stacking positive and negative electrodes in units of a predetermined size with a separator and are then wound. Due to its spirally wound structure, the aforementioned wound electrode assembly is suitable for mounting in cylindrical secondary batteries, whereas it is disadvantageous in terms of space utilization for prismatic or pouch-type secondary batteries. The aforementioned stacked electrode assembly allows for size adjustment when cutting the electrodes and separator, making it easy to obtain a rectangular shape that fits the case; however, it has the disadvantage of a relatively complex manufacturing process and relative vulnerability to external shock. The above stack-and-fold type combines the advantages of the winding type and the stack type by manufacturing bicells and / or halfcells into unit cells of an appropriate size, arranging the unit cells with spacing on a folding separator, and then separating the folding separator to manufacture an electrode assembly.

[0007] The aforementioned pouch-type secondary battery is recently gaining attention as a power source for electric or hybrid vehicles due to its advantages of low manufacturing costs, high energy density, and ease of configuring large-capacity battery packs through series or parallel connections.

[0008] Figure 1 is an exploded perspective view of a typical pouch-type secondary battery.

[0009] As illustrated in FIG. 1, a general pouch-type secondary battery comprises an electrode assembly (1), a plurality of electrode tabs (11, 12) extending from the electrode assembly (1), electrode leads (13, 14) welded and coupled to the electrode tabs (12, 12), and a pouch outer material (2) having a storage portion (21) formed therein for accommodating the electrode assembly (1).

[0010] This pouch-type secondary battery has a structure in which an electrode assembly (1) connected to plate-shaped electrode leads (13, 14) is sealed together with an electrolyte in a pouch outer material (2). A portion of the electrode leads (13, 14) is exposed to the outside of the pouch outer material (2), and the exposed electrode leads (13, 14) are used to electrically connect to a device in which the secondary battery is mounted or to electrically connect the secondary batteries to each other.

[0011] Ultrasonic welding is a technique used when welding the electrode leads (13, 14) to the electrode tabs (11, 12). It refers to a solid-state welding method that generates ultrasonic vibrations of 10,000 Hz to 75,000 Hz and applies the ultrasonic vibrations locally between metals to weld without melting the workpiece. It has the advantage of having a good heat-affected zone (HAZ) and being suitable for welding thin metal foils.

[0012] Figure 2 is a diagram showing the ultrasonical welding of electrode leads to the electrode tabs of a pouch-type secondary battery.

[0013] As shown in FIG. 2, when ultrasonic vibration is applied by an ultrasonic welder (31, 32) while the electrode tabs (11, 12) and electrode leads (13, 14) are in contact with each other, frictional heat is generated along with vibration at the contact surface of the electrode tabs (11, 12) and electrode leads (13, 14), and the electrode tabs (11, 12) and electrode leads (13, 14) are welded together without melting.

[0014] The anode structure, consisting of an anode tab (11) and an anode lead (13), is mainly made of aluminum, and the cathode structure, consisting of a cathode tab (12) and a cathode lead (14), is mainly made of copper or nickel-plated copper. Due to these dual materials, the conditions of the ultrasonic welding process also differ.

[0015] In particular, aluminum anode structures are susceptible to temperature and pressure, which can cause partial fracture in over-welded areas, leading to welding defects. Additionally, foreign matter generated during ultrasonic welding can cause ignition and explosion in secondary batteries. While defects resulting from the overall fracture of the weld area can be identified visually, partial welding defects at specific points are very difficult to detect with the naked eye.

[0016] In this regard, as a method for checking the ultrasonic welding status, conventional technology involved placing pressure-sensitive paper over the workpiece and performing a vision inspection on the point where the color of the pressure-sensitive paper appeared to verify the welding status and any defects of the workpiece.

[0017] However, the above method finds the optimal value through a trial-and-error approach by visually checking the results shown only as color differences on pressure paper and arbitrarily adjusting process conditions; therefore, it is impossible to finely adjust ultrasonic process conditions where various variables exist. In addition, since problems such as unwelded or overwelded welds that cannot be checked visually may occur, the reliability is not high.

[0018] In addition, the "Welding Quality Judgment Device and Method" in Korean Patent Publication No. 2011-0118437 (Publication Date: October 31, 2011) discloses a technology for detecting current and voltage output from a welding machine, processing them into current waveform signals and voltage waveform signals, and then comparing them with a pre-stored waveform signal corresponding to welding to determine welding quality. However, this technology is used for arc welding, which involves applying current directly to a metal material from a welding machine to heat and melt the metal material and bond the two materials by rearranging their atomic bonds, and has limitations in that it is not suitable for ultrasonic welding. Prior art literature

[0020] Korean Patent Publication No. 2011-0118437 The problem to be solved

[0021] The objective of the present invention is to solve the aforementioned conventional problems by providing an ultrasonic welding quality monitoring system and method using artificial intelligence, which can monitor the ultrasonic welding state by collecting the ultrasonic welding shape generated by a single welding of numerous ultrasonic welding devices and the pattern of the voltage signal generated therefrom, and then using artificial intelligence that has undergone prior learning using a Deep Neural Network (DNN).

[0022] Another objective of the present invention is to provide an ultrasonic welding quality monitoring system and method using artificial intelligence that can inspect ultrasonic welding quality in a non-destructive manner by detecting a voltage signal from vibrations generated during ultrasonic welding. means of solving the problem

[0024] As a means to achieve the above-mentioned purpose, in one embodiment, the ultrasonic welding quality monitoring system according to the present invention comprises: an ultrasonic welding device for ultrasonically welding electrode tabs and electrode leads of a secondary battery, which is a workpiece; a vibration sensing unit installed around a booster of the ultrasonic welding device to convert the vibration of the booster into an electrical signal and output it; a voltage sensing unit that senses the voltage of the electrical signal input from the vibration sensing unit at a sampling frequency and outputs it as a voltage signal; and an artificial intelligence unit that collects the ultrasonic welding patterns of numerous ultrasonic welding devices and the patterns of voltage signals generated during a single welding pass, and then performs prior learning using a Deep Neural Network (DNN) to distinguish the ultrasonic welding pattern from the pattern of the voltage signal input from the voltage sensing unit.

[0025] In a specific embodiment, the present invention selects the sampling frequency from a range of 40 to 60KHz, for example, using 50KHz.

[0026] In one embodiment, the present invention further comprises: a voltage signal grouping unit that groups voltage signals input from the voltage sensing unit into a certain unit to form a voltage signal group; an RMS calculation unit that calculates and outputs a Root Mean Square (RMS) value for each voltage signal group input from the voltage signal grouping unit; and a control unit connected to the RMS calculation unit.

[0027] In a specific embodiment, the present invention forms 25 voltage signal groups by grouping 25,000 voltage signals, which are generated by ultrasonic welding once for 0.5 seconds by an ultrasonic device and sensed at a sampling frequency of 50KHz, into groups of 1,000.

[0028] In one embodiment, the present invention comprises a control unit that selects a representative value by extracting a peak value from among the RMS values ​​input from the RMS calculation unit, and then determines a pass / fail result using the ultrasonic welding type input from the artificial intelligence unit and the representative value.

[0029] In one embodiment, the present invention further includes an input interface for a user to input a command; and an output interface for outputting a pass / fail judgment result.

[0030] In another embodiment, the ultrasonic welding quality monitoring method according to the present invention comprises: a step of sensing the vibration of a booster for a single welding of an ultrasonic welding device and converting it into an electrical signal; a step of sensing the voltage of the electrical signal at a sampling frequency and outputting it as a voltage signal; a step of classifying the pattern of the voltage signal; and a step of collecting the ultrasonic welding shape of numerous ultrasonic welding devices and the pattern of the voltage signal generated during a single welding, and then distinguishing the ultrasonic welding shape from the pattern of the voltage signal through prior learning using a Deep Neural Network (DNN).

[0031] In a specific embodiment, the present invention selects the sampling frequency from a range of 40 to 60KHz, for example, using 50KHz.

[0032] In another embodiment, the present invention further includes the step of forming a voltage signal group by grouping voltage signals generated by one welding of the ultrasonic welding device and sensed at a sampling frequency into a certain unit.

[0033] In a specific embodiment, the present invention, in the step of forming the voltage signal group, forms 25 voltage signal groups by grouping 25,000 voltage signals, which are generated by ultrasonic welding once for 0.5 seconds by an ultrasonic device and sensed at a sampling frequency of 50KHz, into groups of 1,000.

[0034] In another embodiment, the present invention further includes the step of calculating a Root Mean Square (RMS) value for each voltage signal group and converting it into an RMS value.

[0035] In another embodiment, the present invention further includes the step of selecting a representative value by extracting a peak value from among the root mean square (RMS) values.

[0036] In another embodiment, the present invention further includes the step of determining good or bad from the representative value. Effects of the invention

[0038] This invention has the effect of being able to monitor the ultrasonic welding state by using artificial intelligence that has undergone prior learning using a deep neural network (DNN) after collecting the ultrasonic welding shape generated by a single welding of numerous ultrasonic welding devices and the pattern of the voltage signal generated therefrom, and to inspect the ultrasonic welding quality in a non-destructive manner by detecting the voltage signal from the vibration generated during ultrasonic welding. Brief explanation of the drawing

[0040] Figure 1 is an exploded perspective view of a typical pouch-type secondary battery. Figure 2 is a diagram showing the ultrasonical welding of electrode leads to the electrode tabs of a pouch-type secondary battery. FIG. 3 is a configuration diagram of an ultrasonic welding quality monitoring system using artificial intelligence according to one embodiment of the present invention. FIG. 4 is a flowchart of the operation of an ultrasonic welding quality monitoring method using artificial intelligence according to another embodiment of the present invention. Figure 5 is a graph showing the pattern of voltage signals for the conditions generated during ultrasonic welding. Specific details for implementing the invention

[0041] Hereinafter, preferred embodiments of the invention will be described in detail with reference to the accompanying drawings so that a person skilled in the art can easily practice the invention. The purposes, functions, effects, and other purposes, features, and operational advantages of the invention, including the purpose, function, and effect, will become clearer through the description of the preferred embodiments.

[0042] For reference, the embodiments disclosed herein are merely the most preferred embodiments selected and presented among various possible embodiments to aid the understanding of those skilled in the art, and the technical concept of the invention is not necessarily limited or restricted only by the presented embodiments, and various changes, additions, and modifications including equivalents and substitutions are possible within the scope of the technical concept of the invention.

[0043] Furthermore, expressions of terms or words used in the specification and claims of this application are defined based on the principle that the inventor may appropriately define the concept of the terms to best describe his invention, and should not be interpreted as being limited only to their ordinary or dictionary meanings, but must be interpreted in a meaning and concept consistent with the technical spirit of the present invention. For example, singular expressions include plural expressions unless the context clearly indicates otherwise, expressions regarding direction are set based on the position shown in the drawings for convenience of explanation, and expressions such as "connected" or "connected" include not only direct connection or connection but also connection or connection mediated by other intermediate components. Additionally, expressions such as "~part" include units realized using hardware, units realized using software, units realized using both hardware and software, etc., and one unit may be realized using one or more hardware or software, and two or more units may be realized using one hardware or software.

[0045] First embodiment

[0046] FIG. 3 is a configuration diagram of an ultrasonic welding quality monitoring system using artificial intelligence according to one embodiment of the present invention.

[0047] As illustrated in FIG. 3, the configuration of an ultrasonic welding quality monitoring system using artificial intelligence according to one embodiment of the present invention comprises: an ultrasonic welding device (40) for ultrasonically welding electrode tabs and electrode leads of a secondary battery, which is a workpiece; a vibration sensing unit (50) installed around a booster (43) of the ultrasonic welding device (40) to convert the vibration of the booster (43) into an electrical signal and output it; a voltage sensing unit (60) that senses the voltage of the electrical signal input from the vibration sensing unit (50) at a sampling frequency of 50KHz and outputs it as a voltage signal; an artificial intelligence unit (70) that collects the ultrasonic welding patterns of numerous ultrasonic welding devices and the patterns of voltage signals generated during a single welding pass, and then performs prior learning using a Deep Neural Network (DNN) to distinguish the ultrasonic welding patterns from the patterns of voltage signals input from the voltage sensing unit (60); and a voltage signal that groups the voltage signals input from the voltage sensing unit (60) into a certain unit. It comprises a voltage signal grouping unit (80) that forms a group, an RMS calculation unit (90) that calculates and outputs the Root Mean Square (RMS) value of each voltage signal group input from the voltage signal grouping unit (80), an input interface (100) for a user to input a command, a control unit (110) connected to the voltage sensing unit (60), the artificial intelligence unit (70), the RMS calculation unit (90), and the input interface (100), and an output interface (120) connected to the control unit (110).

[0048] The above ultrasonic welding device (40) comprises an ultrasonic oscillation unit (41) that generates an ultrasonic oscillation signal of 20KHz or higher, an ultrasonic vibrator (42) that converts electrical energy into mechanical energy by vibrating according to the ultrasonic oscillation signal of the ultrasonic oscillation unit (41), a booster (43) that amplifies the vibration of the ultrasonic vibrator (42), and a horn (44) that performs ultrasonic welding by applying the ultrasonic vibration received from the booster (43) to an electrode tab and electrode lead placed on an anvil.

[0050] Second embodiment

[0051] FIG. 4 is a flowchart of the operation of an ultrasonic welding quality monitoring method using artificial intelligence according to another embodiment of the present invention.

[0052] As illustrated in FIG. 4, the configuration of a method for monitoring ultrasonic welding quality using artificial intelligence according to another embodiment of the present invention comprises: a step (S10) of sensing the vibration of a booster for a single welding operation of an ultrasonic welding device and converting it into an electrical signal; a step (S20) of sensing the voltage of the electrical signal at a sampling frequency of 50KHz and outputting it as a voltage signal; a step (S30) of classifying the pattern of the voltage signal; a step (S40) of collecting the ultrasonic welding forms of numerous ultrasonic welding devices and the patterns of voltage signals generated during a single welding operation therefrom, and then distinguishing the ultrasonic welding forms from the patterns of the voltage signals through prior learning using a Deep Neural Network (DNN); a step (S50) of forming voltage signal groups by grouping the voltage signals generated by the single welding operation of the ultrasonic welding device and sensed at the sampling frequency into a certain unit; a step (S60) of calculating a Root Mean Square (RMS) value for each voltage signal group and converting it into an RMS value; and the The method comprises a step (S70) of selecting a representative value by extracting a peak value from among the RMS values, and a step (S80) of determining pass or fail based on the representative value.

[0053] The operation of the ultrasonic welding quality monitoring system and method using artificial intelligence according to the embodiments of the present invention, based on the above-described configuration, is as follows.

[0054] When an ultrasonic oscillation signal of 20KHz or higher is generated from the ultrasonic oscillation unit (41) of the ultrasonic welding device (40) and applied to the ultrasonic vibrator (42), the ultrasonic vibrator (42) vibrates according to the ultrasonic oscillation signal, converting electrical energy into mechanical energy and transmitting it to the booster (43). The booster (43) amplifies the vibration of the ultrasonic vibrator (42) and transmits it to the horn (44), and the horn (44) applies the ultrasonic vibration received from the booster (43) to the electrode tab and electrode lead placed on the anvil, thereby performing ultrasonic welding.

[0055] In this process, a vibration sensing unit (50) installed around the booster (43) of the ultrasonic welding device (40) converts the vibration of the booster (43) for one welding of the ultrasonic welding device (40) into an electrical signal and outputs it to a voltage sensing unit (60).

[0056] The voltage sensing unit (60) senses the voltage of the electrical signal input from the vibration sensing unit (50) at a sampling frequency of 50KHz and outputs it as a voltage signal (S20).

[0057] The artificial intelligence unit (70) classifies the pattern of the voltage signal input from the voltage sensing unit (60) (S30), and then distinguishes the ultrasonic welding type from the pattern of the voltage signal (S40).

[0058] The above artificial intelligence unit (60) collects the ultrasonic welding patterns of numerous ultrasonic welding devices and the patterns of voltage signals generated during one welding cycle, and then performs prior learning using a Deep Neural Network (DNN), so that it can distinguish the ultrasonic welding patterns from the patterns of voltage signals.

[0059] Figure 5 is a graph showing the voltage signal patterns for conditions generated during ultrasonic welding, and shows the voltage waveforms for the Knurl Pattern Wear Condition, Normal Condition, Horn Tilt Condition, and Bolt Loosen Condition.

[0060] Meanwhile, the voltage signal grouping unit (80) forms 25 voltage signal groups by grouping 25,000 voltage signals, which are generated by ultrasonic welding once for 0.5 seconds by the ultrasonic device (40) and sensed at a sampling frequency of 50KHz, into a certain unit, for example, 1,000 units (S50).

[0061] The RMS calculation unit (90) calculates a Root Mean Square (RMS) value for each of the 25 voltage signal groups input from the voltage signal grouping unit (80), converts it into an RMS value, and outputs it to the control unit (110) (S60).

[0062] The root mean square (RMS) is calculated by finding the average of the squared values ​​of the target signal and finding the square root of the average value in order to determine the characteristics of the target signal, such as an AC voltage signal, and is approximately 70.7% of the maximum value of the target signal.

[0063] The control unit (110) selects a representative value by extracting a peak value from among 25 root mean square (RMS) values ​​input from the RMS calculation unit (90) (S70), then determines whether the product is good or bad using the ultrasonic welding type input from the artificial intelligence unit (70) and the representative value, and then displays the result to the user through the output interface (120) (S80). Explanation of the symbols

[0065] 40 : Ultrasonic welding device 41: Ultrasonic oscillator 42: Ultrasonic transducer 43 : Booster 44 : Soul 50: Vibration sensing unit 60: Voltage sensing unit 70 : Artificial Intelligence Department 80: Voltage signal grouping section 90 : RMS Calculation Section 100 : Input Interface 110 : Control unit 120 : Output interface

Claims

Claim 1 An ultrasonic welding device for ultrasonically welding electrode tabs and electrode leads of a secondary battery, which is a workpiece; a vibration sensing unit installed around a booster of the ultrasonic welding device to convert the vibration of the booster into an electrical signal and output it; a voltage sensing unit that senses the voltage of the electrical signal input from the vibration sensing unit at a sampling frequency selected in the range of 40 to 60 kHz and outputs it as a voltage signal; an artificial intelligence unit that collects the ultrasonic welding shape of the ultrasonic welding device and the pattern of the voltage signal generated during a single welding pass, and then performs prior learning using a deep neural network (DNN) to distinguish the ultrasonic welding shape from the pattern of the voltage signal input from the voltage sensing unit; a voltage signal grouping unit that groups the voltage signals input from the voltage sensing unit into a certain unit to form a voltage signal group; and an RMS calculation unit that calculates and outputs the Root Mean Square (RMS) value of each voltage signal group input from the voltage signal grouping unit. An ultrasonic welding quality monitoring system using artificial intelligence, comprising a control unit that selects a representative value by extracting a peak value from among the RMS values ​​input from the RMS calculation unit, and then determines pass / fail status using the ultrasonic welding type input from the artificial intelligence unit and the representative value. Claim 2 An ultrasonic welding quality monitoring system using artificial intelligence according to claim 1, further comprising: an input interface for a user to input commands; and an output interface for outputting a pass / fail judgment result. Claim 3 A method for monitoring ultrasonic welding quality using artificial intelligence, comprising: a step of sensing the vibration of a booster for a single welding operation of an ultrasonic welding device and converting it into an electrical signal; a step of sensing the voltage of the electrical signal at a sampling frequency selected in the range of 40 to 60 kHz and outputting it as a voltage signal; a step of classifying the pattern of the voltage signal; a step of collecting the ultrasonic welding shape of the ultrasonic welding device and the pattern of the voltage signal generated during a single welding operation, and then distinguishing the ultrasonic welding shape from the pattern of the voltage signal through prior learning using a deep neural network (DNN); a step of grouping the voltage signals generated by the single welding operation of the ultrasonic welding device and sensed at the sampling frequency into a certain unit to form a voltage signal group; a step of calculating a Root Mean Square (RMS) value for each voltage signal group and converting it into an RMS value; a step of selecting a representative value by extracting a peak value from the Root Mean Square (RMS) values; and a step of determining pass or fail based on the representative value. Claim 4 delete Claim 5 delete Claim 6 delete Claim 7 delete Claim 8 delete Claim 9 delete Claim 10 delete Claim 11 delete

Images