LG Electronics announced on the 27th that it has developed AI technology that reduces the quality prediction time of finished products by up to 99% compared to existing methods.
Previously, predicting quality before the production of prototypes took about 3 to 8 hours, but LG Electronics explained that with the new AI technology, it can be completed within 3 minutes without any separate simulations.
Compared to AI technology based on deep learning algorithms, the AI training time for analysis has been reduced by more than 95%, and memory usage has decreased to one-tenth while accuracy has improved by more than 15%.
The convenience for product developers has also been enhanced. By simply inputting 3D drawing information, quality prediction results can be displayed accurately and quickly without a separate process to align the drawing coordinates.
By applying technology to compress and lighten the input data, the AI can learn detailed information more quickly and accurately. Another advantage is that the analysis results can be implemented in a 3D form close to the actual product, allowing developers to review them intuitively.
LG Electronics plans to sequentially apply the AI technology to the design of components used in home appliances such as refrigerators and washing machines, as well as in TVs and automotive products, prioritizing its utilization to enhance its manufacturing capabilities.
Applying the technology to smart factory solutions is expected to help strengthen the competitiveness of the solution sales and consulting business, which began in earnest last year.
Currently, the smart factory solutions are centered on the design, monitoring, and operation of production systems, big data and generative AI-based equipment and process management, industrial safety, quality inspection, and industrial robots, and an AX solution that can enhance product development and production element competitiveness is also in preparation.
Since last year, LG Electronics has been conducting joint research with Professor Park Chan-young's team from the Korea Advanced Institute of Science and Technology (KAIST) to develop verification technology aimed at improving the efficiency of the product design process.
Recently, plans have been made to submit related papers to the International Conference on Machine Learning, a globally recognized top-level conference in the AI field, to obtain academic validation of the technology.
Jeong Dae-hwa, president of the LG Electronics Production Technology Institute, expected that the AI-based verification technology will not only shorten the product development cycle but also significantly enhance development efficiency.