Prediction-based peer-to-peer energy transaction market design for smart grids

Abstract

Predicting energy costs and energy generation in traditional power grids is a complex endeavour, primarily due to the imperative of maintaining supply–demand equilibrium for upholding the stability of the power system. Predicting energy costs and generation includes the highly variable nature of energy prices and the need to analyze large amounts of data from various sources. We propose an auto-executable blockchain-based peer-to-peer(P2P) energy transaction market with a Long short-term memory (LSTM) neural network to address challenges in the energy sector. LSTM can handle non-linear relationships and can effectively model and predict the complex patterns that arise in energy markets. Our energy market enables accurate energy consumption and production prediction, facilitates decentralized and sustainable energy trading, and balances supply and demand via smart contracts. A secure and efficient blockchain-based P2P energy transaction market is designed to facilitate transactions in smart grids. The integrated system is a novel contribution that can improve energy systems’ efficiency, reliability, and sustainability. Our experiments show that our proposed energy market can reduce the electricity buying budget by 17.68% compared to the traditional market and 41.36% compared to the open loop look ahead dispatch method for 30 microgrids and the utility.