Chang, C.-Y.C.-Y.ChangKo, K.-S.K.-S.KoGuo, S.-J.S.-J.GuoHung, S.-S.S.-S.HungLin, Y.-T.Y.-T.LinSY-JYE GUO2021-02-042021-02-042020Chang, C.-Y.;Ko, K.-S.;Guo, S.-J.;Hung, S.-S.;Lin, Y.-T.https://www.scopus.com/inward/record.url?eid=2-s2.0-85084186577&partnerID=40&md5=b3efb28afad92baddc21c34d97c8e5e6https://scholars.lib.ntu.edu.tw/handle/123456789/547363The integrated application of the Internet of Things and artificial intelligence (AIoT) is key during the developmental process from a smart home to smart city and the design of sensors for collecting various types of data is the foundation for establishing the entire AIoT. In this study, we placed our self-designed carbon monoxide (CO) sensor into a 1:10 ratio acrylic house model and simulated three types of CO hazard scenarios. The results comparing 10 cases were used to establish an innovative CO Multi-Forecasting Model (CMFM). The CMFM is suitable for application in the semi-supervised learning - based AIoT. In addition to having a 3-7 times better safety warning time compared to that of commercially available CO sensors, our CO sensor also possesses a health warning function for indoor air quality. © 2020 Taiwan Academic Network Management Committee. All rights reserved.Artificial intelligence; Carbon monoxide poisoning; Prediction; Sensor; Smart home[SDGs]SDG3[SDGs]SDG11Air quality; Carbon monoxide; Indoor air pollution; Semi-supervised learning; CO sensors; Forecasting modeling; Hazard scenarios; Health and safety; Indoor air quality; Integrated applications; Smart homes; Warning time; Automationjournal article10.3966/1607926420200121010232-s2.0-85084186577WOS:000513926800026