Research On Data Security Sharing And Privacy Protection Based On SFedChain

With the deep integration of the internet and the energy industry,new types of energy internet technologies and models have emerged.The widespread deployment of distributed energy storage devices has generated a massive amount of user data.This data is shared among various power generators and distributed energy storage networks to achieve collaborative control and comprehensive analysis,effectively enhancing the clustering and intelligence of distributed energy storage networks.However,the security of data sharing and privacy protection has become a major challenge in the joint modeling and analysis of distributed energy storage networks.In the data transmission process,there are high risks of leakage,which seriously impact the secure sharing of energy storage data.Additionally,security risks such as sensitive information leakage and privacy infringement resulting from centralized data processing pose significant threats to data privacy protection.Therefore,researching methods for secure data sharing and privacy protection of energy storage data in distributed energy storage networks is of great significance in ensuring the security of the energy internet.The paper proposes a data security sharing and privacy protection model in the context of distributed energy storage networks.This model allows for secure and reliable distributed collaborative training of machine learning models without the need to share local data.The research focuses on implementing a secure data sharing method for energy storage data using the SFedChain model,which combines blockchain and federated learning.Additionally,improvements are made to the SFedChain model to address data privacy protection issues arising from improper management of heterogeneous energy storage data in distributed energy storage networks.The specific tasks conducted in this study are as follows:(1)A distributed multi-party data sharing and collaborative training model called SFedChain is proposed.This model effectively reduces the risk of data leakage during the process of joint data modeling.By allowing multiple data holders to collaborate on training models without disclosing their original data,secure data sharing is achieved.(2)Efficient blockchain consensus algorithms and incentive mechanisms suitable for SFedChain are proposed to ensure the honesty,trustworthiness,and improved processing efficiency of the participants in collaborative modeling.To address the security risks,such as data privacy leakage,caused by the centralized processing approach in traditional distributed energy storage networks with diverse data sources,a data privacy protection method based on the improved SFedChain is further proposed.This method leverages deep reinforcement learning algorithms to select subsets of heterogeneous energy storage nodes in the distributed energy storage network,aiming to intelligently choose a subset of devices to participate in collaborative modeling and protect the data privacy of each energy storage node in the network.(3)The paper successfully implements the proposed SFedChain model and the improved SFedChain model.It evaluates the proposed models in terms of training accuracy and training time using benchmark datasets.The effectiveness of the methods is verified through a series of experiments.Additionally,the security and robustness of the proposed approach are demonstrated through simulated attack experiments.