The Research On Smart Grid Data-encryption And Distribution Security Considering Privacy

With the growth of DG penetration rate and the improvement of the electricity price mechanism,the complexity and uncertainty of distribution network are rising rapidly,causing more security problems to distribution network FTU(Feeder Terminal Unit)and smart meters,such as false data injection and electricity stealing.In this context,safety of power data transmission,prevention of false data injection,and detection of electricity stealing will be important conditions for the security ensurance of distribution network,the fairness of power transactions,and the economy of the distribution network.Focus on FTU and smart meters,this paper studies the problems of data tampering on FTU and electricity stealing on smart meter in distribution network.First,the calculation of the Distribution Local Marginal Price(DLMP)and the profit-driven attack behavior of proconsumers are studied to assess the risk of data tampering in the distribution network.Then,the encryption scheme of FTU data and the detection scheme of electricity stealing on smart meter are studied to maintain the security of power data.The main research work of this paper is as follows:(1)A calculation model for DLMP is established based on extreme learning machine to quickly calculate DLMP.Experiment on test set proves that the DLMP calculation model proposed in this paper is both efficient and accurate,with root mean square error 0.0004 on the test set.(2)A false data injection attack decision model based on deep reinforcement learning(DQN)for smart distribution network proconsumers is established.First,the DLMP-based profit model of the producer and consumer is established.On this basis,combined with the DQN model,the attack decision model of the proconsumer is established to analyze the attack intention driven by profit of the proconsumer.Afterwards,data tampering risk of the distribution network is evaluated according to proconsumer’s attack intention.An example is used to verify the effectiveness of the attack decision model and risk evaluation.The example shows that after the pro consumers tamper with FTU data through false data injection attacks,they will obtain about 10% of the excess profits;By resisting attacks during the high-risk period based on the risk evaluation,the excess profits of distribution network proconsumers will be reduced by about 35 %,and 60% will be reduced if attacks are rejected on high-risk node by the risk evaluation.(3)A electricity data encryption scheme based on Paillier encryption algorithm and chaos theory is proposed for FTU.First,theories and concepts of Paillier encryption algorithm and chaos theory are introduced.On this basis,a corresponding power data encryption scheme is established to prevent data from being tampered with.Examples show that the proposed encryption algorithm can effectively encrypt and decrypt FTU power data with effectiveness and security,which can prevent FTU electricity data from tampering.(4)A detection model for electricity stealing on smart meter based on FOA and convolutional neural network is Established.First,electricity stealing problem and its characteristic are introduced.With the combine of FOA and CNN,a detection model for electricity stealing based on historical power consumption is proposed.Examples based on the Pecan Street database in the United States have proved the effectiveness of the model.With 96.3% accuracy on test set,which is 6.7% and 12.8% higher than that of ELM and neural networks under the same conditions,respectively.Proving better ability and practical significance.