Search

As power systems are undergoing a significant transformation with more uncertainties, less inertia and closer to operation limits, there is increasing risk of large outages. Thus, there is an imperative need to enhance grid emergency control to maintain system reliability and security. Towards this end, great progress has been made in developing deep reinforcement learning (DRL) based grid control solutions in recent years. However, existing DRL-based solutions have two main limitations: 1) they cannot handle well with a wide range of grid operation conditions, system parameters, and contingencies; 2) they generally lack the ability to fast adapt to new grid operation conditions, system parameters, and contingencies, limiting their applicability for real-world applications. In this paper, we mitigate these limitations by developing a novel deep meta-reinforcement learning (DMRL) algorithm. The DMRL combines the meta strategy optimization together with DRL, and trains policies modulated by a latent space that can quickly adapt to new scenarios. We test the developed DMRL algorithm on the IEEE 300-bus system. We demonstrate fast adaptation of the meta-trained DRL polices with latent variables to new operating conditions and scenarios using the proposed method, which achieves superior performance compared to the state-of-the-art DRL and model predictive control (MPC) methods. [ABSTRACT FROM AUTHOR]

In this work we propose a novel data-driven, real-time power system voltage control method based on the physics-informed guided meta evolutionary strategy (ES). The main objective is to quickly provide an adaptive control strategy to mitigate the fault-induced delayed voltage recovery (FIDVR) problem. Reinforcement learning methods have been developed for the same or similar challenging control problems, but they suffer from training inefficiency and lack of robustness for “corner or unseen” scenarios. On the other hand, extensive physical knowledge has been developed in power systems but little has been leveraged in learning-based approaches. To address these challenges, we introduce the trainable action mask technique for flexibly embedding physical knowledge into RL models to rule out unnecessary or unfavorable actions, and achieve notable improvements in sample efficiency, control performance and robustness. Furthermore, our method leverages past learning experience to derive surrogate gradient to guide and accelerate the exploration process in training. Case studies on the IEEE 300-bus system and comparisons with other state-of-the-art benchmark methods demonstrate effectiveness and advantages of our method. [ABSTRACT FROM AUTHOR]

In view of the complex traffic flows, spatial interactions within a city exhibit the properties of dynamics, connectivity, and repeatability. This paper aims at mining spatial–temporal movement patterns from massive taxi trajectory data for discovering the inherent travel flow information within the urban system. Similar to the role of ocean circulation in a marine system, identifying the frequent paths and cycles of the travel flows within a city would be critical for understanding the principles behind the travel flow surfaces. Thus, we propose a multi-level method for the discovery of movement paths by incorporating the techniques of network analysis and association rules. Specifically, the proposed method begins by constructing a directed network on the subdivision of the study region, in which the node with geolocation represents the corresponding cell and the edge with weight represents the travel flow between neighboring cells. The method then adopts an extended label propagation clustering algorithm to identify frequent paths and cycles on the flow network within a specific time interval. Finally, to extract frequent paths during the whole time period, we also develop an association rules mining algorithm by modeling the edges as items and the paths in each time span as transactions. Experiment results demonstrate that our framework is able to effectively mine movement patterns in taxi trajectory data. Our results are expected to provide an avenue for further research, such as transportation planning and urban structure analysis. [ABSTRACT FROM AUTHOR]

With the rapid growth of Sentinel-1 synthetic aperture radar (SAR) data, how to exploit Sentinel-1 imagery and achieve effective and robust marine surveillance are crucial problems. In this paper, we present the OpenSARShip, a dataset dedicated to Sentinel-1 ship interpretation. The OpenSARShip, providing 11 346 SAR ship chips integrated with automatic identification system messages, owes five essential properties: specificality, large scale, diversity, reliability, and public availability. These properties make sure that the OpenSARShip achieves its objectives. The first is to provide researchers a benchmark dataset to develop applicable and adaptive ship interpretation algorithms and push the performance ceilings of data analysis. The other is to provide a dataset for performing application-oriented quality assessment for Sentinel-1 imagery, which can boost their applications in a targeting way. The construction and the organization of the OpenSARShip are discussed, which show the inside of the dataset and ensure the essential properties. The elaborate geometric and scattering analyses, the benchmark for classification, and the imagery applicability assessment by using the OpenSARShip all demonstrate the applicability and potential of the dataset. [ABSTRACT FROM PUBLISHER]