Training Convolutional and Auto-Encoder Neural-Networks Based on Deep Learning Approach to Designing PID-Controller for Micro Grid Systems

The main purpose of this study is evaluating and classifying the design control methods for the microgrids systems to maintain stability and load variations by adjusting the controllerparameters especially stand-alone operation. In normal operation, distributed generation units provide power quality control. In different distributed generation, there is a big challenge in controlling the system. In facing with the challenge, two methods PID and MPC have been introduced and compared with each other. The PID controller consists of three filter, proportional, integrate, differential. Using the powerful efficiency of Deep learning methods, the introduced and compared with each other. The PID controller consists of three filter, proportional, integrate, differential. Using the powerful efficiency of Deep learning methods, the map between PID controller input and output can be easily learned via supervising learning. . In this paper, two Neural Networks and training of each of them have been studied which are the two networks, Convolutional Neural Network and Auto Encoder. In this paper, Training of CNN and AE networks based on deep learning approach are presented.

Training Convolutional and Auto-Encoder Neural-Networks Based on Deep Learning Approach to Designing PID-Controller for Micro Grid Systems Keywords:

Deep Learning , Micro Grid Systems , CNN and AE , PID , Training Neural-Network