Integration Of Statistic Analysis And Causal Analysis In Transient Stability Quantization Assessment

With the depletion of fossil fuels and the real threat of global climate change,the new energy revolution must be enforced.Promoting the engineering construction of the ultra high voltage(UHV),smart gird(SG)and energy interconnection(EI)in China is significant for the clean power delivery,which may speed up the energy revolution.More non-linear electric elements are introduced due to the engineering construction of the UHV,SG and EI in China,besides,the system interconnection ability is enhanced further.On one hand,the‘curse’of dimensionality may be further highlighted while carrying out transient stability assessment(TSA).On the other hand,the fault spread speed and range must be increased sharply.What’s more,because of the increasingly frequent natural disasters,high permeability new energy integration,and the more mature power market mechanism,the system operational risks and its uncertainty may be influenced deeply.Hence,the current TSA softwear as well as the stability defense system applied in engineering are facing with challenge.In premise of assuring the transient stability assessment and control(TSA&C)precision,the reduction of computational burden makes sense for the safe,steady and reliable operation of the next generation of bulk power system.For TSA,the existing research is mostly developing the case screening technology only based on statistic analysis.Therefore,computational burden for TSA can be reduced to some extent on the premise that detailed analysis must be executed for the unstable cases.However,related research results nearly haven’t been applied in engineering systems widely.While for generator-tripping control,there is little research about relevant adaptive rapid algorithm.Hence,it is urgent to break the contradiction between the accuracy and speed of TSA&C through the integration of statistic analysis and causal analysis.This dissertation trys to balance the precision and rapidness of TSA&C from the point of algorithm design.Comprehensive,thorough and detailed studies with respect to theoretical analysis,algorithm design and experimental verification,are carried out.And some important conclusions and independent-innovative achievements are made and obtained.1.Clarifying that the source of the contradiction between the accuracy and speed of transient stability quantitative analysis is the objective time-varying factors.And putting forward the design ideas of case screening and case sorting frameworks,which contribute to the balance of accuracy and speed of TSA:The essence of transient stability quantitative analysis,which is the data mining of relevant data sources(e.g.electric parameters),is elaborated.Besides,the electromagnetic power(P_e)-power angle(δ)curve(i.e.P_e(δ)function)is important for the quantification.Based on the formula derivation and theoretical analysis of Hamiltonian and Non-Hamiltonian system respectively,the evolution mechanism of P_e(δ)function due to the consideration of time-varying factors is reveled.To be specific,with the gradually consideration of time-varying factors,the P_e(δ)function evolves from permanent to time-varying,and leading to the contradiction between calculation accuracy and speed.So,indexes which represent the time-varying factors can be served as one of the characteristic time-varying variables for the design of case screening and sorting frameworks.2.Putting forward the method of quantifying the time-varying factors rapidly:The difference between static extended equal area criterion(SEEAC)and dynamic extended equal area criterion(DEEAC),which are the first two stages of EEAC algorithm’s development process,in considering of time-varying factors is elaborated.And from the viewpoint of time-varying factors’overall impact on stability margin as well as its impact on kinetic energy increasing and decressing stages respectively,the information of the difference between SEEAC and DEEAC in performing TSA is extracted,and then the indexes which reflect the time-varying factors are obtained with tiny computational burden.3.Designing the case screening framework based on indexes reflecting time-varying factors:It is revealed that strict case screening is a deadlock,and the solution of the deadlock is put forward.That is,for each case,its quantitative time-varying indexes can reflect the credibility of its approximate stability margin obtained by SEEAC or DEEAC.Then the case with both high enough approximate stability margin and credibility level can be filtered.Several rules for screening out the risk-free cases are designed based on above causal analysis,and the threshold values of these rules are optimized through statistical analysis.What’s more,rules that can filter out what other rules cannot do are selected.And through ranking the selected rules according to their recognition performances from high to low in order,the case screening framework can be finally formed.Its excellent performance is verified by simulations of 9 Chinese regional power systems under various operating conditions.4.Designing the case sorting framework based on indexes reflecting time-varying factors:Satisfying the accuracy and efficiency of TSA simultaneously by only screening out risk-free cases are still facing with challenge.The overall design thinking of case sorting framework is then put forward,that is,each case can be identified into one of the following five categories,namely stable,suspected stable,marginal,suspected unstable and unstable,and only for cases recognized as the marginal category must carry out detailed calculation in any situation.Sorting rules corresponding to each category are set combining the quantitative time-varying indexes,the approximate stability margins,and the fault messages.Then the case sorting framework can be finally built through the selection of rules with unique recognition performance,and ranking them based on each’s priority and recognition effectiveness.Based on simulations of the same test cases,comparing with the case screening framework,the case sorting framework’s advantage is verified.5.Putting forward the method of quantifying the unstable mode variability rapidly:The key point for promoting case sorting performance is extracting new characteristic time-varying variable,which can reflect transient stability characteristics.Based on causal analysis,it reveals that,for unstable cases,the stronger its time-varying degree is,the larger the sensitivity of its unstable mode with respect to scenario parameters(e.g.fault-clearing-timeτ)becomes.Then,utilizing the disturbed trajectories under a sufficiently largeτobtained by large-step Taylor series expansions,the index which reflects unstable mode variability can be quantified rapidly.6.Designing the new case sorting framework based on indexes reflecting time-varying factors and unstable mode variability:Since unstable mode variability index can also reflect the time-varying factors,it can then mark the credibility of the approximate stability margin obtained by SEEAC or DEEAC.The overall design thinking of the new case sorting framework is put forward.To be specific,whole cases can be sorted into three categories,namely,stable,indeterminacy and unstable.It is expected to identify sufficiently more actural stable and unstable cases into stable and unstable categories,and then only the rest cases need to carry out detailed calculation.Based on above causal analysis,new sorting rules corresponding to each category are set with the unstable mode variability index and the previous indexes.Similarily,through statistical analysis,the threshold values of new rules are optimized.And then combining the new and original one,rules which show unique screening performance are selected.Therefore,the case sorting framework is finally formed through ranking the selected one according to their recognization performance from high to low in order.Based on simulations of the same test cases,comparing with the original case sorting framework,the new case sorting framework’s advantage is verified.7.Designing an adaptive algorithm for rapidly optimizing the generator-tripping control:The fast optimization of transient stability emergency control(TSEC)strategy is a key factor of online safeguard for power systems.An engineering applicable TSEC algorithm has to obtain the disturbed trajectories of the system by step-by-step numerical integration,and then do quantized stability knowledge mining.Theoretical analysis and simulation verification reveal that the premise is to identify the dominant mode correctly,and then the emergency control decision can be optimized according to the cost performance of these actions.In order to ensure the accuracy of analysis,both integral step and quantitative knowledge extraction step have to be small enough.Besides,based on causal analysis,it can be learnt that,if the unstable mode is insensitive to the change of fault parameters(e.g.fault clearing timeτ),the time-varying degree of this case is most like weak.For and only for such cases,large-step Taylor series expansions can be taken to replace small-step numerical integration and search for the optimal result.Based on indexes reflecting time-varying factors and unstable mode variability,each test case can be sorted into one of the two categories,i.e.that using large-step Taylor series expansions and that using small-step numerical integration,to get the trajectories,and then do optimized generator-tripping control decision-searching.The adaptive algorithm for rapidly optimizing the generator-tripping control can be finally designed through further combining statistical analysis.The excellent performance of this proposed adaptive algorithm is verified by simulations of the same test cases.