人工智慧物聯網於大學教室室內空氣品質決策管理與應用之研究

A Study of Applying Artificial Intelligence in University-Classroom Indoor Air Quality and Decision Management

本研究運用人工智慧物聯網技術並使用空氣盒子(Air Box)蒐集「環境特徵」變數因子，將該因子作為類神經網路之輸入變數，接著以物件導向程式語言(Python)與數據挖掘工具(IBM SPSS modeler 18.0)建立預測性模型，再採用XGBoost決策樹演算法作為實驗方法，進行模擬建築新風系統模型的準確度。研究結果發現，採用XGBoost決策樹演算法模型準確值平均可高達97.36%以上，證明透過導入智慧感測及人工智慧等感測分析技術結合空氣盒子及建築新風系統，可建構出優化的教學空間空氣品質決策管理運作模式，除可維持教學空間良好室內空氣品質外，同時亦可達到運用人工智慧物聯網技術落實建築室內空氣品質自主管理之目的。

This study utilizes Internet of Things technology with Air Box to gather environmental characteristics factors as input variables to a neural network, which applies the object-oriented programming language (Python) and data mining tools (IBM SPSS modeler 18.0) to establish a predictive model. The results found that the accuracy of the model by the XGBoost decision tree algorithm is up to 97.36%. It indicates that an optimized classroom with a ''Quality Management Model'', an intelligent fresh air system, could be constructed by smart sensing devices, artificial intelligence, and sensing data analysis technologies. In addition to maintaining good indoor air quality within the teaching spaces, it also achieves the purpose of independent indoor air quality management by placing artificial intelligence.