Research On Safe And Low-carbon Scheduling Of Energy Internet Under The Background Of "dual Carbon" Based On Machine Learnin

With the rapid development of the world economy and the continuous burning of fossil energy,the increasing demand for electricity in more areas,along with the consumption of fossil fuels in the process of generating electricity.In this process,the emission of carbon dioxide and other substances has caused certain harm to the environment,the shortage of energy is more serious.Therefore,the continuous increase in the permeability of wind power,photovoltaic and other new energy becomes the development trend of the future power.Due to the randomness and intermittent of new energy power generation,the power system faces the risk of uncertainty,and it becomes necessary to study the energy storage system.The power system has developed into the system of information physical fusion.The attack mode of the power system has changed from physical attack to physical network fusion.The research on its security is urgent.Therefore,the research on low-carbon scheduling and security of energy Internet under the background of "dual carbon" based on artificial intelligence becomes very significant.In this paper,virtual energy storage and pumped storage are combined into generalized energy storage,and demand response constitutes virtual energy storage to expand the power regulation range of the original energy storage system.At the same time,this paper proposes a day-ahead economic dispatch model with generalized energy storage,which is a coordinated optimization of generalized energy storage,wind power and thermal power units,and establishes a reinforcement learning framework to solve this problem by introducing a proximal strategy optimization algorithm in the deep reinforcement learning algorithm.The experiment verifies the effectiveness of the proximal strategy optimization algorithm,and generalized energy storage dispatch can effectively reduce wind abandonment phenomenon.In addition,on the basis of day-ahead economic dispatch,the false data injection attack is introduced,the attack vector model is constructed,the attack vector is quickly solved by GUROBI solver,the predicted load data is changed,and day-ahead economic dispatch is carried out.The experimental results once again verify that the generalized energy storage dispatch can effectively solve the problem of wind abandonment.Finally,from 24 hours economic dispatch to one hour economic dispatch,the false data injection attack is applied to the bus load in the grid topology structure,the mathematical model of vector generation is established,and a simple and efficient algorithm is proposed to solve the problem of the attack vector generation,and the optimal of the algorithm is proved theoretically.After obtaining the false data injection attack vector,economic dispatch is carried out.Experimental results show that the false data injection attack vector solution algorithm can obtain the optimal solution,economic dispatch after injection attacks produces line power overflow,and the detection mechanism in this paper can accurately detect whether the system is under malicious attacks.