Deep Learning-driven Transient Stability Assessment And Emergency Control Of Microgrids

Through effective integration of distributed energy resources（DERs）such as wind,solar energy,and storage energy,microgrid is an important way to promote the consumption of DERs,improve energy utilization and power supply reliability.However,as a typical power electronic power system,microgrid has the characteristics of diverse control strategies,rich types of DERs,low inertia and weak anti-interference ability.As a result,it will face the problems of safe and stable operation,such as equipment burning,unexpected disconnection,system collapse,etc.At present,many documents have deeply analyzed the safe and stable operation of distributed single-DER grid-connected system or isomorphic dual-DERs system,and proposed methods and strategies at the mechanism level,analysis level and control level.The actual microgrid usually contains multiple types of DERs.The existing theoretical basis and analysis methods do not consider the impact of control m ode switching,source-source coupling interaction and other factors on the transient response of the microgrid when multiple DERs are connected.The traditional means used are difficult to support the online safe and stable operation of the microgrid,which limits the large-scale promotion of microgrid technology.For this reason,this paper focuses on the safe and stable operation of microgrid with multi-machine access under large disturbance,and carries out research work around the transient operation characteristics of microgrid,online transient stability assessment（TSA）,and online emergency control（EC）.The main achievements are as follows:（1）Aiming at the problem of unclear transient operation mechanism of microgrid considering source-source interaction.First,the transient interaction models of current-controlled DER（CC-DER）parallel system,voltage-controlled DER（VC-DER）parallel system and heterogeneous DER parallel system are established.Then the full-time scale mathematical expressions of fault current of various DER in homogeneous/heterogeneous systems are derived.The analytical solution and numerical calculation methods are used to quantitatively describe the transient operation characteristics with source-source interaction,and the main factors affecting the transient operation characteristics are analyzed and summarized.It is found that in the homogeneous system,the source interaction is an important factor affecting the operation characteristics of DERs.In heterogeneous systems,the control parameters of CC-DER have a greater impact on the positive sequence fault current of the VC-DER,while the VC-DER has little impact on the transient operation characteristics of the CC-DER.Relevant conclusions provide a theoretical basis for the design of reliable operation of microgrid.（2）Aiming at the system-level TSA of multi-VC-DER system under the current limiting strategy.First,a unified transient instability analysis model of VC-DER is estabilished.Then the influence mechanism of the P-f loop,Q-V loop,load fluctuation and fault depth on the transient stability of VC-DER under the current limiting strategy is explored.Based on the transient stability mechanism,the principle of selecting the input features of the multi-VC-DER system with"system dynamic parameters as the main part and system steady-state parameters as the auxiliary part"is proposed,and the online TSA method of the multi-VC-DER system based on the deep feedforward neural network is proposed.The proposed meth od does not need to be equipped with additional measuring equipment,and can realize rapid prediction of system-level stability at the moment of fault clearing.It can ensure accurate prediction of system-level stability under current limiting,DER control mode change,fault change,load fluctuation and other operation conditions.（3）Aiming at the system-level TSA of microgrid considering heterogeneous DER interaction.First,the coupling mechanism of transient instability of heterogeneous DER system is revealed.It is found that the stability of heterogeneous DER system is mainly determined by its own dynamic term and dynamic interaction term.Through qualitative analysis of the interactive dynamic term,it is concluded that the coupling angle composed of the fault current injection angle of the CC-DER and the power angle of the VC-DER is the key factor affecting the system transient stability.Based on the theoretical results,a general selection method of key input features for TSA of microgrid with various control strategies is proposed.This method makes the assessment model independent of the size of microgrid topology and can avoid the curse of dimensionality.Next,based on the deep belief network,a microgrid TSA method suitable for small samples is proposed,which solves the online evaluation problem of microgrid under the interaction of heterogeneous DERs.（4）Aiming at the DER level refined TSA problem of microgrid considering spatial-temporal dynamic characteristics.It is found that the transient instability region of DERs is spatially and dynamically correlated with the change of fault location,and the time segment of DER transient trajactory is dynamically correlated in temporal dimension.Thus,the impact of spatial-temporal dynamic characteristics between DERs on the transient stability cannot be ignored.Based on this,an adaptive capture and modeling method of spatial-temporal dynamic correlation based on attention mechanism is proposed to extract the hidden spatial-temporal dynamic characteristics of DER.On this basis,a microgrid refinement TSA model based on spatial-temporal graph convolution neural network is proposed to realize microgrid instability DER identification and visualization,DER instability degree change trend prediction.Furthur,DER level transient stability ranking,instability time and other indicators can be given to provide a knowledge basis for microgrid refinement EC.（5）Aiming at the robust stabilization problem of online transient stability EC for microgrid.First,the transient stabilization conditions of VC-DER under three transient instability conditions are analyzed,and it is concluded that P₀,J₀ and D_P are the keys to ensure the fast regression stability of DER.Based on theoretical research,a microgrid EC strategy based on parameter flexible rolling control is proposed.This strategy also includes a more rapid and novel control end criterion,which coordinates the transient stability at the DER level and the system level.In addition,based on the improved spatial-temporal graph convolution network,a microgrid output power prediction model considering the impact of rolling time window control sequence is designed to provide support for the online and rapid implementation of the proposed EC strategy.The verification results show that the strategy does not need to be equipped with additional measuring equipment,and can ensure the fast,robust,and low control cost stabilization of the microgrid in any case of instability.This paper has carried out systematic work on the safe and stable operation of microgrid in the event of large disturbance,focusing on the three aspects of transient operation mechanism,TSA,and transient stability EC.It has carried out the transient operation characteristic analysis considering the source-source interaction,the multi-VC-DER system-level prediction considering the current limiting,the microgrid system-level prediction considering the interaction of heterogeneous DER,the DER level refined prediction considering the spatial-temporal dynamic correlations and online transient stability EC of microgrid based on parameter rolling control.This paper provides a systematic theoretical basis and technical support for the safe and stable operation of microgrid under large disturbances.