融合机器学习与区块链的电能交易系统的研究与设计

Research and Design of Electric Energy Trading System Integrating Machine Learning and Blockchain

现今区块链驱动的能源交易平台逐渐付诸实践,其实际应用场景中的各类安全问题也随之暴露。其中,最令人们关注的问题就是参与交易的用户信息追溯和用户身份认证问题。在传统能源交易系统中,用户认证依赖于第三方,例如账户登陆与人脸检测等,但该类方式不但具有信息泄露的风险,还无法将用户信息准确同步于区块的加密封装中。因此,提出一种应用机器学习提取笔迹特征以进行用户真实信息认证的方式,通过用户进行实时笔迹签名,对用户的笔触、笔迹、内容、风格等进行特征提取。在不直接利用生物信息的前提下,获取能够充分代表用户信息与身份的数据特征,且达到高于传统2D人脸识别方法的准确率。同时,提出一种将该类特征整合入区块链封装流程的方法,使得整体交易系统具有用户数据准确且身份信息可追溯的优点。最后对提出的系统进行性能测试,测试结果表明,所提的方法除了能提高整体系统的安全性以外,还提升电力交易系统的交易效率。

As the blockchain-driven energy transaction platform become prevalently used in real live,many practical problems have come along.One of the most typical problems is the identity verification issue.In traditional methodologies,identification is independent with the core transaction system,making the system threated by data leakage and the inability to incorporate identification features into the block encodings.Thus,this paper propose a method which implements machine learning methods to attain the chirography features such as speed,content and pressure variations for identification uses.Such method successfully bypasses the biological information leakage while achieving a higher accuracy than average 2 D face recognition methods.This article also proposes a method to integrate such features into the blockchain packaging process,so that the overall transaction system has the advantages of accurate user data and traceability of identity information.