http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Real-time Fault Diagnosis for Train Doors on Edge AI Device
Angela Caliwag,Donguk Kwon,Wansu Lim 한국통신학회 2021 한국통신학회 학술대회논문집 Vol.2021 No.2
The reliability and safety of train systems is necessary as it transports massive number of passengers during its operation. To improve the reliability and safety of train systems, several researchers proposed fault diagnosis methods to identify the type of fault that is present in a system. Most of the fault diagnosis methods available in the literature were implemented and tested using simulations on desktop computers. In this paper we implement a real-time fault diagnosis algorithm on an edge AI device to detect the fault on doors of train. Since the signals of the doors of train has minimal deviations, a deep learning-based algorithm is adopted to extract feature and classify the type of fault that is present in the system.
Angela Caliwag,임완수 한국통신학회 2021 ICT Express Vol.7 No.4
The error in state of charge estimation using the combined models is usually attributable to the statistical model. In this study, a least square algorithm is utilized to optimize and increase the state of charge estimation accuracy. Specifically, the vector autoregressive moving average statistical model is optimized using the least square algorithm. The results presented in this paper show that the proposed method is effective in eliminating the estimation and measurement noise using the conventional statistical method and in optimizing the SoC estimation and forecasting. The optimization of the statistical model increases the SoC estimation and forecasting accuracy by 59.11%.
AWS Data Visualization using DynamoDB and Lambda
Ej Miguel Francisco Caliwag,Angela Caliwag,Donguk Kwon,Wansu Lim 한국통신학회 2021 한국통신학회 학술대회논문집 Vol.2021 No.2
AWS IoT has been one of the extremely powerful AWS service offers. AWS IoT has the ability to process more than billions and trillions of devices and messages respectively and can process and routes to endpoints (i.e. AWS DynamoDB, Aws Lambda). In this paper, we proposed a method for temperature data sensing using DHT22 sensor connected to Raspberry Pi 4B that interacts to AWS IoT which will allow us to connect to AWS DynamoDB for storing, AWS Lambda for serverless trigger in AWS services, AWS Quicksight for visualization and AWS Machine Learning for further data processing.
Continuous Emotion Recognition on the Edge
Ej Miguel Francisco C. Caliwag(이제이),Henar Mike O. Canilang(헤나르),Angela C. Caliwag(안젤라),Wansu Lim(임완수) 한국통신학회 2021 한국통신학회 학술대회논문집 Vol.2021 No.6
Traditional Edge AI implementation methods requires high cost computing machines / device. Aside from these high cost devices, most implementation often requires complex AI models. In this study, a real-time implementation for Edge AI devices is implemented to address the implementation cost and complexity constraints. The Arduino Nano 33 BLE sense is used as an Edge AI Device for this implementation. An open source machine learning platform is used to develop an embedded machine learning implementation which addresses the high implementation and computational complexity of edge AI implementation.