The steady-state security region of the integrated energy system (IES) is a useful tool for rapid security assessment and security evaluation of the integrated energy system. A complete characterization of the IES steady-state security region is derived. The derivation is under the nonlinear and non-convex AC power flow model, the gas flow model, and the uncertainty of renewable energy without any linear simplification. Then a novel and robust computation scheme to compute the IES steady-state security region is proposed and shown to eliminate the estimation errors of the existing methods. It is shown that the IES steady-state security region is non-convex and composed of several disjoint components. Numerical studies are conducted to show that the IES steady-state security regions obtained from the existing methods can be inaccurate, while the proposed characterization gives an exact estimation of the IES steady-state security region that significantly improves the previously proposed methods.

In order to effectively implement remedial reactions to mitigate the negative impacts of oscillations on a power grid, it is essential for system operators to timely and accurately determine whether an observed oscillation is a natural oscillation or a forced oscillation. Using phasor measurement unit (PMU) data, this paper proposes a residual spectral analysis (RSA) method to distinguish forced oscillations and natural oscillations. The proposed RSA method uses forecasting models with various lead times to forecast the current PMU data based on the past PMU data. The spectra of the forecasting residuals are shown to have different properties when the PMU data have forced or natural oscillations. Support vector machines are applied to the residuals to classify the oscillations. This paper develops an algorithm for implementing the RSA method, and demonstrates its superior performance via extensive simulations over the 48-machine model. Simulations show that it can distinguish the oscillations more accurately than an existing method and work reliably even when the frequency of forced oscillations is close to that of natural oscillations.

In this paper, detectability analysis is studied and extended to guide selection of measurements for the dynamic state estimation (DSE) of a synchronous machine. To make sure that the DSE converges, past studies suggested that enough measurements must be available to make the system observable. In this paper, the convergence condition is relaxed via detectability analysis to reduce the number of required measurements. Because the objective of the DSE is to estimate the current states as time progresses, it is shown that a DSE observer can converge not only for an observable system but also for an unobservable system if the eigenvalues corresponding to the unobservable states are stable. Simulation results using the IEEE 10-machine, 39-bus system show that the unscented-Kalman-filter-based DSE can converge when measurements are chosen such that the unobservable states have stable eigenvalues. In comparison with observability analysis, the proposed application of detectability analysis can reduce the number of measurements required for the existence of a DSE observer.

Accurate quantification of uncertainty in solar photovoltaic (PV) generation forecasts is imperative for the efficient and reliable operation of the power grid. In this paper, a data-driven non-parametric probabilistic method based on the Naïve Bayes (NB) classification algorithm and Dempster–Shafer theory (DST) of evidence is proposed for day-ahead probabilistic PV power forecasting. This NB-DST method extends traditional deterministic solar PV forecasting methods by quantifying the uncertainty of their forecasts by estimating the cumulative distribution functions (CDFs) of their forecast errors and forecast variables. The statistical performance of this method is compared with the analog ensemble method and the persistence ensemble method under three different weather conditions using real-world data. The study results reveal that the proposed NB-DST method coupled with an artificial neural network model outperforms the other methods in that its estimated CDFs have lower spread, higher reliability, and sharper probabilistic forecasts with better accuracy.

In this paper, an online multiperiod power dispatch with renewable uncertainty and storage is presented. The ac power flow equations, voltage limits, and thermal limits are respected to ensure the obtained solution causes no static violation while ensuring network transfer capability. A scenario reduction technique using affinity propagation clustering is developed to reduce the number of scenarios and relieve the computational burden. A three-stage solution methodology is developed to solve the nonconvex constrained optimization problem, in which a two-parameter homotopy-enhanced methodology with a decomposition scheme is developed to reliably and quickly solve the online multiperiod power dispatch problem. The proposed three-stage solution methodology has been evaluated on several testing systems, ranging from a 30-bus system to a 3012-bus system with four time periods, and the numerical results are promising.