Korea Advanced Institute of Science and Technology

Assembly Line

Development of an injection molding production condition inference system based on diffusion model

📅 Date:

✍️ Authors: Joon-Young Kim, Heekyu Kim, Keonwoo Nam

🔖 Topics: Injection molding, process parameter inference, diffusion

🏢 Organizations: Korea Advanced Institute of Science and Technology


Plastic injection molding is a crucial process for the mass production of various products. However, traditional methods for setting production conditions have heavily relied on skilled operators to adjust parameters through trial and error. This approach is not only inefficient but also results in inconsistent quality control. To address these challenges, this study proposes a new machine learning based model that automatically infers process parameters, enabling real time adaptation to external environmental changes. A surrogate model is first developed to learn the relationship between process parameters, environmental variables, and product quality, predicting whether a given set of parameters will result in a good or defective product. Building on this, a diffusion model, a type of deep generative model, was employed to generate diverse sets of process parameters likely to yield defect free products under specific environmental conditions. The proposed diffusion model outperforms existing generative models such as generative adversarial network (GAN) and variational autoencoder (VAE) in both accuracy and diversity of generated parameters. Notably, the diffusion model achieved an error rate of 1.63%, significantly outperforming GAN and VAE, which exhibited error rates of 23.42% and 44.54%, respectively. Additionally, the applicability of the proposed diffusion model was experimentally validated in a real world testbed. Several experiments conducted under various external environmental conditions demonstrated that the quality of the products produced using the process parameters generated by the diffusion model matched the quality predicted by the model. This study introduces a novel approach to improving both the efficiency and quality of injection molding processes and holds promise for broader applications in manufacturing.

Read more at Journal of Manufacturing Systems

New System Monitors EV Batteries

📅 Date:

✍️ Author: Austin Weber

🏭 Vertical: Electrical Equipment

🏢 Organizations: Korea Advanced Institute of Science and Technology


Engineers at the Korea Advanced Institute of Science & Technology (KAIST) have developed a way to precisely determine the health of batteries by only using small amounts of electrical current. Their electrochemical impedance spectroscopy (EIS) technology may improve the stability and performance of high-capacity batteries in electric vehicles. In addition to assessing the state of charge and state of health of batteries, the tool can be used to identify thermal characteristics, chemical or physical changes, predict battery life and determine the causes of failures.

Traditional EIS equipment is expensive and complex, making it difficult to install, operate and maintain. And, due to sensitivity and precision limitations, applying current disturbances of several amperes to a battery can cause significant electrical stress, increasing the risk of battery failure or fire. The KAIST system can precisely measure battery impedance with low current disturbances, minimizing thermal effects and safety issues during the measurement process. It minimizes bulky and costly components, making it easy to integrate into EVs.

Read more at Assembly

Fault Detection and Identification using Bayesian Recurrent Neural Networks

📅 Date:

✍️ Authors: Weike Sun, Antonio R C Paiva, Peng Xu

🔖 Topics: process monitoring, Tennessee Eastman, Bayesian recurrent neural network

🏢 Organizations: Korea Advanced Institute of Science and Technology


In processing and manufacturing industries, there has been a large push to produce higher quality products and ensure maximum efficiency of processes. This requires approaches to effectively detect and resolve disturbances to ensure optimal operations. While the control system can compensate for many types of disturbances, there are changes to the process which it still cannot handle adequately. It is therefore important to further develop monitoring systems to effectively detect and identify those faults such that they can be quickly resolved by operators. In this paper, a novel probabilistic fault detection and identification method is proposed which adopts a newly developed deep learning approach using Bayesian recurrent neural networks (BRNNs) with variational dropout. The BRNN model is general and can model complex nonlinear dynamics. Moreover, compared to traditional statistic-based data-driven fault detection and identification methods, the proposed BRNN-based method yields uncertainty estimates which allow for simultaneous fault detection of chemical processes, direct fault identification, and fault propagation analysis. The outstanding performance of this method is demonstrated and contrasted to (dynamic) principal component analysis, which are widely applied in the industry, in the benchmark Tennessee Eastman process~(TEP) and a real chemical manufacturing dataset.

Read more at arXiv

Statistical Process Monitoring of the Tennessee Eastman Process Using Parallel Autoassociative Neural Networks and a Large Dataset

📅 Date:

✍️ Authors: Seongmin Heo, Jay H Lee

🔖 Topics: process monitoring, Tennessee Eastman

🏢 Organizations: Korea Advanced Institute of Science and Technology


In this article, the statistical process monitoring problem of the Tennessee Eastman process is considered using deep learning techniques. This work is motivated by three limitations of the existing works for such problem. First, although deep learning has been used for process monitoring extensively, in the majority of the existing works, the neural networks were trained in a supervised manner assuming that the normal/fault labels were available. However, this is not always the case in real applications. Thus, in this work, autoassociative neural networks are used, which are trained in an unsupervised fashion. Another limitation is that the typical dataset used for the monitoring of the Tennessee Eastman process is comprised of just a small number of data samples, which can be highly limiting for deep learning. The dataset used in this work is 500-times larger than the typically-used dataset and is large enough for deep learning. Lastly, an alternative neural network architecture, which is called parallel autoassociative neural networks, is proposed to decouple the training of different principal components. The proposed architecture is expected to address the co-adaptation issue of the fully-connected autoassociative neural networks. An extensive case study is designed and performed to evaluate the effects of the following neural network settings: neural network size, type of regularization, training objective function, and training epoch. The results are compared with those obtained using linear principal component analysis, and the advantages and limitations of the parallel autoassociative neural networks are illustrated.

Read more at Processes