Power System Transient Stability Assessment Based On Accident Information

Power system is one of the most complex strongly nonlinear, high-dimensional, layered and distributional great dynamic industrial systems, whose stability assessment and control issues have always been focus and difficulty of electric power research. In recent years, with growth of load, increase of transmission distance, application of various new technologies and control devices, implementation of power market policy, large-scale access of wind power, solar energy and other new energy, putting into operation of mass HVDC and FACTS systems, power system dynamic characteristics, like coherence, instability mode, critical flowgate and so on, become more and more complex, which increases the difficulty of power system assessment and control. In the future, with the advancing of smart grid construction, the utilization of power system equipments will be further enhanced on the existing basis, which will has push the operating conditions of power system ever closer to its stability limits. As a result, the possibility of cascading accident occurring between different AC channels, DC channels, AC/DC channels and sending/receiving grid will be higher and higher. Considering potential failures of relay protection device, mutual influence of multi-infeed and AC grid, unscheduled outage of generators, reactive power support ability of receiving grid, low frequency oscillation, transient stability, voltage stability and other factors, the risk of cascading accident evolving to large-scale blackout in future power system will truly increase. Because transient stability accident is one of the main factors which induces cascading accident evolving to large-scale blackout, fast and accurate assessment of transient stability is very important for effective implementation of preventive control, emergency control and corrective control. Therefore, modern power system stability analysis in urgent need of an effective method to quickly assess the stability status of the system after a accident, which can provide important information support for power system operators to take appropriate, timely and effective control measures.With the development of wide-area measurement technology and artificial intelligence technology, power system transient stability assessment methods based on wide-area measurement information and artificial intelligence algorithm gain wide attention of researchers.In order to make full use of power system accident information provided by wide area measurement system, this paper conducts a study of power system transient stability assessment based on artificial intelligence and energy function algorithm, the main research contents are as follows: (1) This paper proposes a power system transient stability assessment model based on support vector machine(SVM) with combinatorial trajectories inputs. A binary support vector machine classifier with combinatorial trajectories inputs was trained to predict the transient stability status in this model. Besides, a credible area and an incredible area of the classifier were given to improve the practicality of the classifier and a revised strategy was proposed to improve the performance of the SVM classifier in incredible area.(2) This paper proposes individual machine potential ridge criteria by introducing the concept of individual machine energy function and generalized potential energy function to ridge potential method. Taking into account of model differences between energy function and actual grid, constraints to improve reliability of ridge potential criteria are presented based on the actual grid operation features. Above all, a transient stability assessment method based on individual ridge potential criteria using real-time wide-area measurement information is presented.(3) Aiming at the disadvantages of the PEBS(potential energy boundary) method, this paper proposes a high reliable power system transient stability analysis and critical cutset detection criteria named branch potential ridge method based on branch potential energy and potential ridge method. Besides, a simple stability index of branch and system together with a method to detect the stability status and critical cutset of power system based on BPR and branch stability index are proposed. Compared with existing branch potential energy function method, the proposed method is more accurate and simpler. In the case of known parameters of network structure, the method only needs real-time data of bus voltage magnitude and phase angle to do online power transient stability analysis and critical cutset detection. Because the method doesn’t make assumption of two group instability mode, this method can be used in multi-swing instability and multi-group instability mode. Because the method only uses network structure data and bus dynamic fault information, it can also be applied in power system with large-scale wind power and other renewable energy integration.(4) This paper proposes a power system critical cutset detection method named rolling double level hierarchical clustering method. First, this method uses generator variable trajectories and bus variable trajectories to do respectively hierarchical clustering and gains grouping results of generators and buses. Then, the above two hierarchical clustering results are used for mutual verification to improve the reliability of power system critical cutset detection result. Also, in order to reduce the calculation work of distance matrix in hierarchical clustering of bus variable trajectories and guarantee the connectivity of bus group gained from hierarchical clustering, a bus variable distance calculation method considering power system network topology is proposed, which greatly improve the efficiency of proposed critical cutset detection method.(5) Based on the perspective of available transmission capacity (ATC) and critical cutset, this paper proposes a new power system transient stability margin--transient available transmission capacity (TATC). TATC directly uses power to quantitatively measure the degree of power system transient stability, which is convenient for power system operators to fast formulate prevention or emergency control measures to guarantee the safe and stable operation of power system. Besides, in order to achieve online calculation of the proposed TATC, a fast TATC prediction method based on support vector machine(SVM) with initial state accident information inputs is proposed.