Artificial Intelligence Techniques in Power Systems

To reduce the computational effort of the security assessment, at the present time, most energy management systems (EMSs) use one or more security assessment (SA) predictors (such as sensitivity matrices, distribution factors, fast decoupled load flows, or performance indicators) to reduce the number of critical contingencies to be calculated explicitly in real-time. There are many SA predictors available, each making certain assumptions about the network or the operating states, in order to reduce the computation effort in the evaluation. For example, the simplest scalar SA predictors are of the scalar category having the general form [46]:
where PI stands for performance index; h k( u, C l) is a real-valued function of ( u, C l) { w k} are positive weighting coefficients. The notation PI( u, C l) emphasises that the PI is evaluated at the operating point, u, and contingency, C l. This PI is used both for insecure contingency detection and severity ranking. The contingency classification is performed using the following criterion:

where TH is some specified or calculated threshold value. This criterion is used for either thermal loading limit violation checking, or voltage limit violation checking, or voltage stability checking when the PI in eqn. 8.7 represents either a loading index, or a voltage index, or a stability index, respectively.
Although this type of SA predictor is computationally efficient, it may not...