Grid Security Risk Assessment Based On Random Forest Algorithm

Large-scale centralized renewable energy power generation,extreme natural disasters and other uncertain factors have put forward higher requirements for the operation security risk prevention and control ability of power grid.Electric transmission line faults caused by extreme natural disasters with a small probability may sometimes lead to malignant accidents of large-scale power failure,which will bring great harm to economic development and social stability.Reasonable use of the current rapid development of artificial intelligence technology can improve the intelligent level of power grid security risk prevention and control,improve the security risk assessment ability and the early warning speed.Therefore,given that renewable energy can cause the uncertainty of power system operation mode,with the aid of artificial intelligence means such as machine learning,it is important to realize efficient and accurate assessment of the safety and stability of power grid operation risks,which is helpful to take corresponding measures from the perspective of power grid planning and dispatching operation to effectively prevent and control risks.Firstly,the characterization method of power system operation mode considering source side output uncertainty is studied.Taking wind power as an example,the dynamic scenario generation method considering the randomness and volatility of wind power is studied.By identifying the key parameters of the covariance matrix of multiple normal distributions and sampling,a large number of dynamic scenarios conforming to the randomness and volatility of wind power output are generated.By clustering for large dynamic scenarios,this paper realized scenario reduction,acquiring typical scenarios conforming to wind power uncertainty and the probablitty of the typical scenarios.Then this paper obtains the power system operation mode of each deterministic typical scenario by optimizing.Based on the different operation modes,grid security risk assessment of the renewable energy is carried out.Then,because the uncertainty of renewable energy output will lead to the uncertainty of system operation mode,the time domain simulation analysis of a large number of operation modes is too time-consuming to obtain the security control cost,which brings difficulties to the safety risk assessment of power grid operation.By means of artificial intelligence,offline training and online application of samples can improve the efficiency of security risk assessment.In this paper,a large number of off-line samples are generated for learning and testing of stochastic forest machine learning algorithm by kernel call of power system analysis and synthesis program(PSASP).Based on the random forest machine learning algorithm,the following two aspects were studied: first,the limit cutting time of relay protection after the expected failure of the transmission line under different operation modes was studied,and the transient power Angle stability results were predicted online.Secondly,the minimum safety and stability control cost of the transmission line under different operation modes after the expected failure is studied,realizing the accurate prediction of the minimum safety and stability control cost of the transient power Angle instability on line,providing the control cost basis for the calculation of the safety risk assessment.Safety risk assessment based on random forest machine learning algorithm can effectively improve the efficiency of risk assessment,and compare with the prediction accuracy of neural network algorithm,which shows the superiority of random forest machine learning algorithm in prediction accuracySecondly,the fault probability of electric transmission line is quantitatively evaluated when natural disasters are taken into account.By reading static information such as state of power grid equipment,topography and landform along the line,and dynamic forecast information related to meteorological conditions,fault probability of transmission lines under rainstorm-induced landslide,mountain flood and ice damage is calculated respectively by means of membership function in fuzzy mathematics and related fuzzy rule table.According to the probability calculation rule of independent events,the fault probability of electric transmission line under rainstorm disaster and compound natural disaster is further obtained,which provides a basis for the calculation of security risk assessment.Finally,the calculation method of safety risk index is studied.The risk assessment results of transmission lines are obtained by comprehensively considering the factors that affect risk indicators,such as source side output uncertainty,control cost under different operation modes of the system and electric transmission line failure probability caused by natural disasters.The risk values of different transmission lines calculated according to the above methods can be ranked,normalized and clustered.The system safety and stability risk problem with source side output uncertainty is comprehensively and comprehensively evaluated in this paper,providing guidance for further prevention and control.