| With the upgrading of power system,while maintaining high reliability and security,building an efficient,intelligent and multi-functional grid is a challenging task,especially in today’s evolving network threat environment.Therefore,how to ensure the privacy of users’ information and how to ensure that theft can be detected in time is still a challenge for smart grid.Therefore,security and privacy are the key issues in the design and implementation of smart grid,which directly affects the safe and reliable operation of smart grid and users’ satisfaction with power companies.When illegal elements theft energy,legitimate users or power companies will suffer economic losses and even threaten the security of the whole power grid.Firstly,the detection technology of energy theft in smart grid is deeply studied,and the detection of energy theft in smart grid at home and abroad is analyzed and studied.At the same time,the user’s electricity usage information may be exposed when the user data is detected for stealing energy,and the user’s privacy is involved in the energy usage information,which should be protected.Therefore,on the basis of systematic research on security and privacy protection of smart grid,aiming at the emergence of stealing and privacy problems in smart grid,after in-depth research,this paper proposes feasible solutions,namely,the scheme of stealing detection with privacy protection based on state estimation and the scheme of stealing detection with privacy protection based on convolutional neural network,the former is more applicable.In micro-grid,the latter can process huge amounts of energy data.State estimation is a method of estimating the internal state of a dynamic system based on available measurement data.Kalman filter is very important for state estimation.In smart grid,Kalman filter can be used in state estimation algorithm to realize the deviation of user’s power consumption.By comparing the deviation with the actual deviation of user,the user’s power consumption behavior can be analyzed.Through the state estimation,we get the deviation between the estimated value and the actual value of the user.If the deviation exceeds the threshold,we can determine the abnormal power consumption of the user.The data used for detection can be directly the energy consumption,so it is more correct to use the state estimation method to analyze the user’s energy consumption behavior through the obtained energy consumption information,and the simulation results will be more intuitive.At the same time,because of the detection of energy consumption,the actual operation of the power company can be considered in the scheme,which can protect the privacy of energy consumption,so as to protect the personal safety of users.Convolutional neural network(CNN)is a well-known and widely used deep neural network in the field of image recognition,including in the competition of authoritative image networks,the top algorithms in the list mostly come from CNN,such as famous VGG,Resnet and so on.More importantly,CNN algorithm plays an important role in data processing in matrix form.This provides an algorithmic support for the analysis and detection of energy usage data.In this paper,a combined convolutional neural network model is established to detect abnormal energy theft based on the similarity of users’ energy consumption behavior in the same user group.The use of user group data helps overcome the problem of incomplete data,solve the more complex detection of stealing,and ultimately improve the detection accuracy.In addition,the use of homomorphic encryption to achieve privacy protection and data aggregation and efficient smart grid communications,in the premise of protecting user privacy to achieve smart grid scheduling. |
