It is increasingly important to create a healthier indoor environment for office buildings. Accurate and reliable prediction of PM_2.5 concentration can effectively alleviate the delay problem of indoor air quality control system. The rapid development of machine learning has provided a research basis for the indoor air quality system to control the PM_2.5 concentration. One approach is to introduce the CatBoost algorithm based on rank lifting training into the classification and prediction of indoor PM_2.5 concentration. Using actual monitoring data from office building, we consider previous indoor PM_2.5 concentration, indoor temperature, relative humidity, CO₂ concentration, and illumination as input variables, with the output indicating whether indoor PM_2.5 concentration exceeds 25 μg/m³. Based on the CatBoost algorithm, we construct an intelligent classification prediction model for indoor PM_2.5 concentration. The model is evaluated using actual data and compared with the multilayer perceptron (MLP), gradientboosting decision tree (GBDT), logistic regression (LR), decision tree (DT), and k-nearest neighbors (KNN) models. The CatBoost algorithm demonstrates outstanding predictive performance, achieving an impressive area under the ROC curve (AUC) of 0.949 after hyperparameters optimition. Furthermore, when considering the five input variables, the feature importance is ranked as follows: previous indoor PM_2.5 concentration, relative humidity, CO₂, indoor temperature, and illuminance. Through verification, the prediction model based on CatBoost algorithm can accurately predict the indoor PM_2.5 concentration level. The model can be used to predict whether the indoor concentration of PM_2.5 exceeds the standard in advance and guide the air quality control system to regulate.

Building envelopes and indoor environments exhibit thermal inertia, forming a virtual energy storage system in conjunction with the building air conditioner (AC) system. This system represents a current demand response resource for building electricity use. Thus, this study centers on the CatBoost algorithm within machine learning (ML) technology, utilizing the LASSO regression model for feature selection and applying the Optuna framework for hyperparameter optimization (HPO) to develop a cost-optimal control method for minimizing building AC loads. This method addresses the challenges associated with traditional load forecasting and control methods, which are often impacted by environmental temperature, building parameters, and user behavior uncertainties. These methods struggle to accurately capture the complex dynamics and nonlinear relationships of AC operations, making it difficult to devise AC operation and virtual energy storage scheduling strategies effectively. The proposed method was applied and validated using a case study of an office building in Nanjing, China. The prediction results showed coefficient of variation in root mean square error (CV-RMSE) values of 6.4% and 2.2%. Compared with the original operating conditions, the indoor temperature remained within a comfortable range, the AC load was reduced by 5.25%, and the operating energy costs were reduced by 24.94%. These results demonstrate that the proposed method offers improved computational efficiency, enhanced model performance, and economic benefits.