A Comparative Study of Deep Neural Network and Meta-Model Techniques in Behavior Learning of Microgrids

Abstract

Behavior learning of microgrids (MGs) is a necessary and challenging task for multi-MGs cooperation and energy pricing of distribution energy market. With the increasing demand for user privacy, this problem becomes more severe because of much less limited access to device parameters and models behind the Point of Common Coupling (PCC), which hinders conventional model-based power management methods. In this paper, to address this problem, some novel model-free data-driven methods including Deep Neural Network (DNN) and Meta-model techniques, such as Radial Basis Function (RBF), Response Surface Methods (RSM), and Kriging methods are introduced. These methods can predict the behavior of MGs through continuous iterative learning by accessing merely the historical active power measurements at the PCCs as well as public electricity price and weather information behind the PCCs, without full system identification and no prior knowledge on the system. A comparative study has been fully carried out by comparing with the conventional model-based model to better understand their advantages, drawbacks and limitations. The validity and applicability of the proposed methods is verified by numerical experiments. This paper can provide some references for future MGs interactive operation under incomplete information. INDEX TERMS Data-driven method, artificial neural network, meta-model techniques, micro-grids behavior learning. I. NOMENCLATURE SETS AND INDICES P is set of solar irradiance data P ws set of wind speed data P T set of temperature data γ P set of electricity price data K set of random variables for sampling M set of PCC power data ACRONYMS MG Microgrid PCC Point of Common Coupling The associate editor coordinating the review of this manuscript and approving it for publication was Salvatore Favuzza .

Electric Distribution Network With Multi-Microgrids Management Using Surrogate-Assisted Deep Reinforcement Learning Optimization

“…A deep neural network (DNN) trained through deep learning has an impressive capability for learning complex system behavior [60], especially the high uncertainty power system behavior. Moreover, the performance of MG behavior learning using the DNN surrogate model is guaranteed by Xiao et al [61] that it has good adaptability to predict the MG behavior required high-dimension inputs and provides high accuracy for the prediction. Hence, this work develops three MG as a DNN surrogate model.…”

Real-Time Optimal Operation Control of Micro Energy Grid Coupling With Electricity- Thermal -Gas Considering Prosumer Characteristics

“…P2G is dispatched by MEG-S operators, so there is no separate power purchase cost. Therefore, the operation cost of P2G (C p2g MEG-S ) only considers the maintenance cost, as shown in (31).…”

“…Nowadays, the idea and method of machine learning is developing rapidly, and it is gradually used to solve the problems in power system. With the development of computing ability and algorithm, reinforcement learning and deep learning technology have been used to solve renewable energy output forecasting, user side load forecasting and complex system optimization [31]- [35]. In [31] and [32], deep neural network and meta-model techniques are used to learn the behavior of MG and determine the optimal operating schedules of the heat generation equipment.…”

Data-Driven Dynamical Control for Bottom-up Energy Internet System

“…In [29], based on deep neural network techniques, an intelligent multi-MG energy management method is applied to increase the profitability of electricity trading. Furthermore, the deep reinforcement learning (DRL) method has now been regarded as a powerful tool to solve such data-driven optimization issues, and reader can refer to [30], [31] and the references therein for more examples. Compared with the traditional multi-stage optimization methods, DRL has unparalleled advantages in end-to-end training, utilizing global information and continuous control [32].…”