Research On Security Risk Analysis Methods For System Operation In Power System With Wind Penetration

Power system with large-scale wind power penetration has increasing difficulties insystem operation. For the research on risk assessment methods of operation in powersystem with wind penetration, from the operation risk analysis perspective, this paperstudies the vulnerability of power system with wind power, reasonable nodes for integratedwind farm, operation risk analysis using probabilistic load flow and the application inscreening dispatch modes and optimal dispatch with stochastic constraints. The complexnetwork and self-organized criticality theory, probabilistic analysis method andmulti-objective evolutionary algorithm are adopted and studied further. The main researchand findings are as follows.This paper presents a method based on the decomposition of electrical betweennessand vulnerability analysis to solve the wind farm location and wind power integratedcapacity range for power planning. This paper decomposes the node electrical betweennessfrom its electric principle, to find the key branches which frequently influence weak nodes.Then, according to the node pair influencing the key branches greatly, the wind farmlocation and integrated capacity range are studied. This paper presents that the wind poweris not suitable for integrating into the source side because of avoiding stochastic windpower to influence the security of key lines and nodes. The method has fast computationspeed and can screen the right located node from an overall and macroscopic perspective.This paper studies the probabilistic load flow improved by modified Latin hypercubesampling (LHS) with evolution algorithm. For solving non-positive definite correlationmatrix effectively, an improved median Latin hypercube sampling method withEvolutionary Algorithm was proposed. Latin hypercube important sampling (LHIS)technique was presented to consider the tail of distribution, and LHIS with evolutionalgorithm was proposed to control correlation. The method can deal with generator reactive power constraint, local correlation of input random variables and the non-positive definitecorrelation matrix of initial sample. The method is flexible and has high accuracy forsolving the off-line probabilistic problem in power system planning and operation with agood application value.This paper studies the operation risk analysis method in power system with wind farmbased on probabilistic load flow, which using improved LHS can be used to prove therationality of wind integrated nodes and screen the operation modes satisfying securityprobabilistic constraints. The cases show that the proposed method can provide moreoverall information for further risk management.This paper studies the environmental and economic dispatch (EED) problem based onoperation risk analysis and builds the deterministic optimization model of the thermalpower system, deterministic optimization model of system with wind penetration and itsstochastic program model. A hyper-cloning technique is proposed to improve thedistribution of solutions obtained by non-dominated neighbor immune algorithm (NNIA).The limitation of the approaches solving different models in the time scale is discussed.This paper builds the multi-objective deterministic optimization and stochastic programmodels considering risk and adopts the security risk analysis method to screen theoperation mode in the latter. The deterministic model of EED is suitable for onlinecalculation. The stochastic program model of EED is based on scenarios set, andinevitablely samples many scenarios making long time computation, suitable for off-linecalculation with no requirement for time. This paper proves that NNIA and non-dominatedsorting differential evolution (NSDE) both have high efficiency and good solution quality.Based on the NNIA, the hyper-cloning technique has better solution distribution thanproportional cloning.This paper studies the evolution rule and convergence problem in the theory basis ofmulti-objective evolutionary algorithm (MOEA), and presents the explanation for thespatial evolution character of the MOEA using self-organized criticality. This paper summarizes the basic strategies playing a key and necessary role in improving the MOEA.The spatial evolution character is modeled by the statistical property of crowding distance,which displays scale-free feature and a power-law distribution. Besides, this paper presentsthe imprecise of the convergence index, and presents the limitation of the small-worldtopology in the application of MOEA. The proposed viewpoint provides a new perspectiveand explanation for clearly understanding the MOEA, distinguishing the strategies playingkey roles and judging the truth of effectiveness of algorithms.