A Transfer and Deep Learning-Based Method for Online Frequency Stability Assessment and Control

Abstract

Fast and accurate prediction and control of power system dynamic frequency after disturbance is essential to enhance power system stability. Machine learning methods have great potential in harnessing data for online application with accurate predictions. This paper proposes a two-stage novel transfer and deep learning-based method to predict power system dynamic frequency after disturbance and provide optimal event-based load shedding strategy to maintain system frequency. The proposed deep learning model combines convolutional neural network (CNN) and long short-term memory (LSTM) network to harness both spatial and temporal measurements in the input data, through a four-dimensional (4-D) tensor input construction process including, 1) capture system network topology information and critical measurements from different time intervals; 2) compute a multi-dimensional electric distance matrix and reduce to a 2-D plane which can describe the system nodal distribution; 3) construct 3-D tensors based on state variables at different sample times; and 4) integrate into 4-D tensor inputs. Moreover, a transfer learning process is employed to overcome the challenge of insufficient data and operating condition changes in real power systems for new prediction tasks. Simulation results in IEEE 118-bus system verify that the CNN-LSTM method not only greatly improves the timeliness of online frequency control, but also presents good accuracy and effectiveness. Test cases in the New England 39-bus system and the South Carolina 500-bus system validate that the transfer learning process can provide accurate results even with insufficient training data.

Frequency Stability Prediction of Power Systems Using Vision Transformer and Copula Entropy

“…The synchronization factor between node j and node k is represented as: where V and δ denote the voltage amplitude and phase angle difference, respectively, and B jk and G jk denote the transfer impedance. Therefore, the voltage amplitude and phase angle of each bus [ 32 ] should be added to the original features. Furthermore, the power imbalance for a generator is defined as: where P m and P e represent the mechanical power and electrical power of the generator, respectively.…”

“…f N stands for the system operational frequency. Referring to Equation (16), the electrical power values of the generators [ 32 ] are also selected as original input features. Note that the electrical power values of the nonsynchronous generators also might be related to frequency stability, according to [ 55 , 56 , 57 ].…”

“…Each DL method mentioned above has corresponding advantages. Therefore, based on these characteristics, some scholars have developed multi-model combinations, such as CNN-LSTM [ 32 ], CNN-GRU [ 33 ], and GNN-LSTM [ 34 ] models, to further mine data from various aspects of information for prediction purposes. It is noteworthy that none of the former studies focused on transformer models in the research field of power system stability.…”

“…Frequency prediction indicators can be divided into two kinds: frequency curve prediction [ 32 , 45 ] and frequency characteristics prediction [ 46 , 47 ]. However, the above frequency indicators only distinguish between frequency security and insecurity.…”

“…Numerical simulations were implemented on a modified New England 39-bus system with PSS/E [ 61 ] to simulate the data acquired by a WAMS and PMUs. To approximate the system behavior [ 27 ], the load levels of the power system were set to 50%, 51%, …, and 100% of the original load levels [ 32 ]. The same ratio was also used to scale the generation power, but extra modifications were provided to ensure that all input data fall within a reasonable range [ 27 ].…”