Search for: All records

In this paper, a data-driven method is proposed for fast cascading outage screening in power systems. The proposed method combines a deep convolutional neural network (deep CNN) and a depth-first search (DFS) algorithm. First, a deep CNN is constructed as a security assessment tool to evaluate system security status based on observable information.With its automatic feature extraction ability and the high generalization, a well-trained deep CNN can produce estimated AC optimal power flow (ACOPF) results for various uncertain operation scenarios, i.e., fluctuated load and system topology change, in a nearly computation-free manner. Second, a scenario tree is built to represent the potential operation scenarios and the associated cascading outages. The DFS algorithm is developed as a fast screening tool to calculate the expected security index value for each cascading outage path along the entire tree, which can be a reference for system operators to take predictive measures against system collapse. The simulation results of applying the proposed deep CNN and the DFS algorithm on standard test cases verify their accuracy, and the computational efficiency is thousands of times faster than the model-based traditional approach, which implies the great potential of the proposed algorithm for online applications. 

Unit Commitment is usually formulated as a Mixed Binary Linear Programming (MBLP) problem. When considering a large number of units, state-of-the-art methods such as branch-and-cut may experience difficulties. To address this, an important but much overlooked direction is formulation transformation since if the problem constraints can be transformed to directly delineate the convex hull in the data pre-processing stage, then a solution can be obtained by using linear programming methods without combinatorial difficulties. In the literature, a few tightened formulations for single units with constant ramp rates were reported without presenting how they were derived. In this paper, a systematic approach is developed to tighten formulations in the data pre-processing stage. The idea is to derive vertices of the convex hull without binary requirements. From them, vertices of the original convex hull can be innovatively obtained. These vertices are converted to tightened constraints, which are then parameterized based on unit parameters for general use, tremendously reducing online computational requirements. By analyzing short-time horizons, e.g., two or three hours, tightened formulations for single units with constant and generation-dependent ramp rates are obtained, beyond what is in the literature. Results demonstrate computational efficiency and solution quality benefits of formulation tightening. 

Unit Commitment is an important problem faced by independent system operators. It is usually formulated as a Mixed Binary Linear Programming (MBLP) problem, and is believed to be NP hard. To solve UC problems efficiently, an idea is through formulation tightening. If constraints can be transformed to directly delineate an MBLP problem’s convex hull during data preprocessing, then the problem can be solved by using linear programming methods. The resulting formulation can be reused for other data sets, tremendously reducing computational requirements. To achieve the above goal, both unit- and system-level constraints are tightened with synergistic combination in this paper. Unit-level constraints are tightened based on existing cuts and novel “constraint-and-vertex conversion” and vertex projection processes. To tighten system-level constraints, selected cuts are applied and some potentially powerful cuts are identified. Numerical results demonstrate the effectiveness of tightening unit- and system-level constraints.