Predicting Failure Cascades in Large Scale Power Systems via the Influence Model Framework
Abstract : Large blackouts in power grids are often the consequence of uncontrolled failure cascades. The ability to predict the failure cascade process in an efficient and accurate manner is important for power system contingency analysis. In this paper, we propose to apply the influence model for the prediction and screening of failure cascades in large scale DC and AC power networks. Then, the trained influence model is applied to some large power grids with thousands of buses and transmission lines. The prediction performance is evaluated in four different aspects, from the global perspective of the overall failure size to the individual information regarding the link failure time.The results show that under limited training samples, the proposed framework is capable of predicting the failure cascade size with a 7% error rate, the final state of links with a 10% error rate, and the failure time within 1 time unit for both the DC and AC model. One major advantage of the proposed method is that it can reduce the computational time of the failure cascade prediction by a few orders of magnitude with limited compromise in accuracy, as compared to the power flow based contingency analysis. Another important feature of the proposed method is that the trained influence parameters can reveal the critical initial contingencies. This information is very helpful for identifying the worst contingency scenarios for system operators.
? In order to further simplify the network model while preserving the basic cascade dynamics, a number of existing works completely ignore the power flow constraints and concentrate on the contagious cascade process in which new failures only happen over adjacent components.
? It is referred to as the “contagion model” which originates from the percolation model.
? In order to construct the influence model that can beused for power system failure cascade predictions, Hines et. al. applied the Monte Carlo method to learn the influences from historical data
? They focused on generating the appropriate distribution of failure cascade sizes that best match existing records.
? In CFS after each failure event, the DC/AC power flow problem is solved using the MATPOWER Toolbox. A detailed oracle of generating synthetic failure cascades is presented in Appendix.
? We can solve these small optimization problems in parallel in practice to further reduce the training time.
? Once have been obtained, their values can be substituted in to form a reduced optimization problem whose decision variables are only from D.
? We choose the objective function f(•) to be the least square error function, which induces the following convex quadratic programming.
• We proposed a hybrid learning framework that can efficiently train the influence model for very large systems.
• The proposed learning framework integrates Monte Carlo method with quadratic programming, and an adaptive threshold selection scheme to quickly train the model for making good predictions.
• One of the future research directions is to predict failure cascades under the AC power flow model. The AC model will not change the underpinning of the proposed hybrid learning scheme.
• In this paper, we propose a hybrid learning framework that can be used to train the influence model from either historical data or synthetic data
? To evaluate prediction performance via the proposed learning framework, we consider two sets of metrics: sample-based metrics and link-based metrics.
? Therefore, evaluating the prediction performance on links with different failure frequencies can deepen our understanding of the model and features in a very detailed manner.
? Therefore, we cannot say that the deterministic prediction performs well for links with failure frequency .
? This tells that we need to combine the miss detection rate lmd and the false alarm rate lfa together to judge and compare the performance.
|