In secondary battery manufacturing, a defective weld between the electrode tab and the electrode lead can raise internal resistance, cause localized heating, and in severe cases lead to an internal short circuit — which is why inspection technology that captures both accuracy and speed at once remains fertile ground for new patents. On February 2, 2026, TechDial Co., Ltd.’s “Weld Inspection Apparatus and Method for Electrode Tab and Electrode Lead” was granted as a patent in Korea (Registration No. KR102920584B1, Publication No. KR20240167335A). Interestingly, one of the prior-art references cited during that patent’s examination was LG Energy Solution’s 2021 application (published 2023), “System and Method for Ultrasonic Weld Quality Monitoring Using Artificial Intelligence” (Publication No. KR20230040581A, currently still under examination). This piece compares, claim element by claim element, two patents that solve the same problem using two different physical signals — X-ray and ultrasound.
① What’s at Stake
Electrode tab-lead weld inspection has long relied on a mix of methods — electrical resistance measurement via radiographic exposure, visual/vision inspection, physical fracture testing, and X-ray CT inspection. Of these, X-ray CT has been criticized for being precise but too time- and cost-intensive to apply as a 100% inspection method for pouch-type batteries. TechDial’s newly registered patent is an attempt to solve this bottleneck by combining microfocus X-ray transmission imaging with deep learning, while LG Energy Solution, around the same time, filed an application for a real-time monitoring method using an entirely different physical signal — ultrasonic vibration. Placing the two companies’ approaches side by side confirms that the rights landscape in this field has not yet settled into a single standard.
② Basic Mechanism
TechDial’s method can be likened to reading depth from two X-ray photographs. A microfocus X-ray tube transmits X-rays through the weld, a high-resolution detector converts the received signal into a digital image, and the resulting 3D weld-spot pattern is then fed into a pre-trained deep learning algorithm to judge whether the weld is defective. It reads as an approach that secures inspection speed by relying on transmission imaging rather than full X-ray CT, while using deep learning to compensate for a level of discriminative power approaching that of 3D imaging.
Source: reconstructed by the editorial team from the claims of KR102920584B1 and KR20230040581A
LG Energy Solution’s method, by comparison, is closer to a stethoscope listening for the tremor in the sound a weld makes. A sensor attached to the booster (the vibration-transmitting component) of the ultrasonic welder converts the vibration generated during welding into an electrical signal; this voltage signal is sampled and compared against patterns learned by a deep neural network (DNN) to monitor weld quality in real time. Its distinguishing feature is that, unlike X-ray, it requires no separate inspection process — judgment happens online, at the very moment the weld is being made.
③ Comparison — Key Configuration of the Two Patents
| Category | TechDial KR102920584B1 | LG Energy Solution KR20230040581A |
|---|---|---|
| Detected physical quantity | X-ray transmission image (3D weld-spot pattern) | Ultrasonic vibration (voltage RMS signal) |
| Core hardware | Microfocus X-ray tube + high-resolution detector | Vibration sensor attached to ultrasonic welder booster |
| Judgment algorithm | Pre-trained deep learning (image-based) | Deep neural network (DNN, voltage-signal pattern-based) |
| Inspection timing | Separate post-weld inspection process | Real-time (online) monitoring during welding |
| Rights status (as of September 2026) | Registered (Active); registered 2026-02-02, expected to expire 2043-05-19 | Published application (Pending); filed 2021-09-16, under examination |
Source: compiled by the editorial team from the Google Patents public records (bibliographic data and claims) of both documents
What this table shows: the two patents diverge clearly in the signal they judge and in the timing of inspection — the relationship looks less like direct competition than a complementary one, where each could sit at a different stage of the process.
④ Patent Perspective
Claim 1 of the TechDial patent (KR102920584B1) is built from four main elements — (a) a microfocus X-ray tube that irradiates the weld with X-rays, (b) an X-ray detector that converts the transmitted light into a digital image signal, (c) a processing unit that extracts the weld-spot pattern from that image, and (d) a weld-judgment unit that judges the extracted pattern using a pre-trained deep learning algorithm. The dependent claims (claims 2–7) further narrow the scope of protection through elements such as reconstructing a 3D image by varying the X-ray irradiation angle, setting a region of interest (ROI), and the composition of the training dataset.
Claim 1 of LG Energy Solution’s application (KR20230040581A) consists of (a) an ultrasonic welding device, (b) a sensing unit that converts the booster’s vibration into an electrical signal, (c) a signal-processing unit that samples the electrical signal, and (d) an artificial-intelligence unit that monitors quality using patterns learned by a DNN.
| Element category | TechDial KR102920584B1 | LG Energy Solution KR20230040581A | Likelihood of scope overlap |
|---|---|---|---|
| Detected physical quantity | X-ray transmission image | Ultrasonic vibration (voltage RMS) | Low — different physical signals |
| Signal-acquisition hardware | X-ray tube + detector | Sensor attached to welder booster | Low |
| Judgment algorithm | Pre-trained deep learning (image input) | DNN (voltage-signal pattern input) | Conceptually similar, but input data differs |
| Inspection timing | Separate post-weld process | Real-time during welding | Low — different process-stage placement |
| Target cell form factor | Described mainly around pouch-type | No form factor specified | Partial overlap possible |
Source: editorial interpretation based on the claim language of both documents (not an infringement opinion)
As this mapping shows, both patents approach the same problem — electrode tab-lead weld inspection — using different physical signals, which supports an assessment that the likelihood of direct scope conflict (infringement) is low. That said, both patents share the higher-level concept of “judgment based on patterns learned via deep learning/DNN,” so if a third-party company were to develop a hybrid inspection system combining both physical signals, it would need to be separately reviewed which side’s dependent claims that system might come into conflict with — which in turn suggests this kind of combined design could actually open up new freedom-to-operate white space.
Design-around scenario (counterargument): If a third-party company sought to develop an optical-camera-based weld inspection system, there is room to design around the core elements of both patents by adopting an optical image — rather than X-ray or ultrasound — as the detected physical quantity. In that case, however, whether it would conflict with separately existing optical-vision-based weld inspection patents (for example, a patent for CCTV-integrated AI vision on a battery electrode line) would need to be reviewed independently. It should be noted that this assessment is strictly a conceptual comparison based on the published claim language, and that an actual freedom-to-operate (FTO) determination is a matter of likelihood that can only be settled by reviewing the specific mode of implementation together with the full limitations of all dependent claims.
⑤ Limits and Outlook
Optimistically, the fact that two different physical approaches to the same problem have each been granted patent protection suggests that technical diversity in this field has not yet reached saturation. But there is a counter-scenario as well — if the industry standard narrows toward combining the two methods (post-weld X-ray inspection and real-time ultrasonic monitoring) into a single integrated inspection system, the commercial value of patents covering each individual method risks becoming dependent on whichever company holds the combined patent.
As a strategic alternative, one option worth examining is securing a patent for a hybrid inspection system that actually combines both technologies — for example, a structure that performs primary screening via ultrasound during welding and reserves precision X-ray re-inspection only for samples where an abnormal signal is detected. Such a combination could raise inspection efficiency without infringing the scope of either prior patent.
Action for IP staff: Confirm which physical signal — X-ray, ultrasonic, or optical vision — the weld inspection equipment your company has adopted or is considering uses, and review, element by element, whether the claims corresponding to that method overlap with the two patents above (or their relevant dependent claims).
Reference: Google Patents KR102920584B1 (published as KR20240167335A), applicant TechDial Co., Ltd., filed 2023-05-19 · registered 2026-02-02 · Google Patents KR20230040581A, applicant LG Energy Solution Co., Ltd., filed 2021-09-16 · published 2023-03-23 · Both documents confirmed directly from the original text on patents.google.com in September 2026.