Research On Winter Personnel Window Opening Behavior Recognition Model Based On Actual Data

Personnel behavior has a significant impact on building energy consumption and is also a key factor causing uncertainty in building energy consumption.With the continuous optimization and improvement of building energy efficiency standards,the airtightness and thermal performance of building envelope structures are continuously enhanced.Indoor personnel have increasingly high requirements for environmental comfort.As one of the most common human behaviors during winter heating,window opening behavior not only improves indoor air quality,but also brings huge energy consumption losses for heating.Therefore,research on the behavior of opening windows in cold regions during winter is of great significance for formulating energy-saving design plans for buildings and achieving intelligent regulation of heating.There is currently limited research on the recognition of window opening behavior among winter personnel in cold regions.This article adopts a combination of actual measurement and research methods to collect data on window opening behavior in real buildings.Based on the actual data,data mining methods such as K-modes clustering algorithm and C4.5 decision tree algorithm are used to conduct in-depth research and mining on winter window opening behavior of urban residential residents in cold regions.The main tasks are as follows:Firstly,a detailed questionnaire survey was conducted on the winter window-opening behavior of residential building occupants in cold regions,which included basic information of occupants,evaluation of indoor thermal and humid environment and air quality environment in winter,characteristics of winter window-opening behavior and drivers of window-opening behavior.The importance of the drivers of window opening behavior in winter was ranked as follows: temperature > air quality > humidity.The characteristics of winter window-opening behavior were also analyzed,and the characteristics of people’s window-opening behavior,including window opening time,window opening length and window opening method,were obtained.Secondly,a representative house in a city in Shandong Province was selected as a typical house for a 60-day test.The actual winter window-opening behavior of the typical house was continuously monitored and four typical winter window-opening behavior patterns were derived using the K-modes clustering algorithm.Then,based on the four typical winter window opening behavior patterns,the indoor temperature and humidity changes were analyzed under different window opening behavior patterns,and the "average temperature drop per hour after window opening","temperature drop in the first hour after window opening" and "temperature drop in the whole process after window opening" were analyzed under each window opening pattern.The "average temperature drop per hour after window opening","temperature drop in the first hour after window opening" and "temperature drop in the whole process after window opening" are quantified and analyzed.And use Design Builder energy consumption simulation software to simulate the energy consumption of typical residential rooms,analyze the impact of different window opening behavior modes on building heating energy consumption,and conclude that winter window opening behavior will cause a large amount of heat loss and increase building heating energy consumption.Finally,using the C4.5 decision tree algorithm,the seven parameters of indoor temperature,indoor relative humidity,15 min indoor temperature change,60 min indoor relative humidity change,outdoor temperature,outdoor relative humidity and hour-by-hour solar incidence angle of each wall surface were selected as input variables based on the research and measurement results,and a winter window opening behavior recognition model based on indoor temperature and humidity and outdoor environment parameters was established.The model is validated by calculating the accuracy and precision of the model.The recognition results are consistent with the trend of the real window opening situation,and the accuracy of the recognition model is 89.6%～94.8%,and the precision rate is84.9%～91.6%.