چكيده به لاتين
Nowadays, due to the high energy demand, modifying the consumption pattern is a very important debate. In our country, due to the relatively warm climate in most areas, the summer season presents many challenges for electricity generation. The purpose of this project is to design a system that modifies the user's consumption pattern and thus reduces the consumption curve amount. This reduces user costs and reduces the load on power grids. The design of this system is done in Python language and in the PyCharm environment and also deep learning models were trained in Google Colab cloud service. In this project, a compatible framework is designed to be implemented on the Raspberry Pi board.
Using deep learning networks and having 3 years of consumption data for network training, user consumption patterns are predicted. For the best performance of the target network, the data set of the target dataset must be prepared and the missing data would be recovered.
After receiving peak power consumption from the power distribution network and removing it from the daily schedule, the system follows two methods (TLM and TM) depending on the type of device (controllable or uncontrollable). The user will notify the system when his interact with device will be finished and may reject the device propose after it has been offered and the service would suggest another device. The TLM method uses the system to network with sensors and execute commands on connected objects.
