With the rapid development of the Internet of Things and artificial intelligence, the intelligent window opening and closing system has become a key component of the modern smart home environment regulation. In view of the many defects presented by the traditional control method in the process of utilizing the switch window system, this study develops a multi-parameter cooperative control algorithm based on feed-forward neural network, which is unique in that it organically combines the principles of physics with the data-driven approach. The physically guided feedforward neural network (PGFNN) architecture we constructed not only enhances the physical interpretability of the system, but also significantly improves its generalization ability in the face of complex environments by cleverly embedding indoor aerodynamic and thermodynamic models. The study shows that the PGFNN control algorithm has significant advantages in the synergistic adjustment of multi-dimensional parameters such as temperature, humidity and air quality, and exceeds the traditional PID control and standard feed-forward neural network control scheme in terms of both control accuracy and response speed. The PGFNN algorithm shows outstanding adaptability and stability when environmental conditions change drastically, and the PGFNN algorithm also performs well in energy utilization, which can effectively reduce the energy consumption of the system while guaranteeing the control effectiveness. This study provides innovative ideas and practical methods for the design and performance optimization of the smart window switching system, which is of substantial significance for improving the control performance and user comfort of the overall smart home system.
