Design of Neural Network Assisted Blockchain based Peer-to-Peer Electrical Energy Trading Platform

Abstract

As traditional power grid is restructuring into smart grid, the concept of peer-to-peer electricity market is becoming a topic of immense interest among the researchers. With this blockchain enabled electricity trading platform has been gaining momentum owing to its capability to provide secure transactions. Work has been reported in recent times employing blockchain technology for implementing decentralized electricity marketing. Most of the work focuses on designing electricity marketing rules only on the smart contract. In the presented work a neural network based solar and wind power prediction model is shown working in tandem with blockchain based simple auction peer-to-peer electricity marketing mechanism. This gives direct advantage to the prosumers to auction their surplus power in day ahead marketing by having right knowledge of solar and wind power prediction of their plant. This reduces risk factor for prosumer paying penalty in case the promised energy delivery is not fulfilled and at the same time load to demand balance can be maintained. A smart contract for peer-to-peer electricity marketing rules has been implemented on Ethereum blockchain platform and deployed on Ropsten Test Network. The neural network based solar and wind power prediction model has been designed on MATLAB R2018b. At the end a demonstration is done considering a scenario of prosumer auctioning solar and wind power to different consumers on blockchain Ropsten test network. This research work does not consider the power distribution network constraints.