Multi-factor indicator of THIC intelligent lighting system with BP neural network

The factors affecting human comfort in office environments are complex, and numerous thermal comfort researches have been conducted by scientists. Among them, The Predicted Mean Vote (PMV) model stands among the most recognized thermal comfort models. However, there are many factors affecting comfort in office places such as illuminance, illuminance contrast, temperature, humidity, location, light color and color temperature. By analyzing the factors influencing the comfort of the office, the most important factors affecting human comfortable in the office are determined, mainly including temperature, humidity, illuminance, color temperature (THIC). By using temperature and humidity monitoring module BME280, GY30 illuminance sensor, TCS34725 full-color temperature sensor and data processing module, the data of temperature, humidity, illuminance and color temperature were collected. After collation, the comfort function of a single factor was obtained. Through a questionnaire survey, combined with analytic hierarchy process and method of two elements contrasting, the comfort model combining temperature, humidity, illumination, color temperature and other multi-factors was established. According to the characteristics of the factors, we trained the model with BP neural network algorithm. This model has a maximum recall of 95.05% and a maximum precision of 71.44%, and the average values of recall and precision after 20 times of model runs are 86.53% and 66.82%. It shows that the model can predict comfort and help control the hardware intelligently in the office.