Support Vector Machine-Based Algorithm for Post-Fault Transient Stability Status Prediction Using Synchronized Measurements

Abstract

The paper first shows that the transient stability status of a power system following a large disturbance such as a fault can be early predicted based on the measured post-fault values of the generator voltages, speeds, or rotor angles. Synchronously sampled values provided by phasor measurement units (PMUs) of the generator voltages, frequencies, or rotor angles collected immediately after clearing a fault are used as inputs to a support vector machines (SVM) classifier which predicts the transient stability status. Studies with the New England 39-bus test system and the Venezuelan power network indicated that faster and more accurate predictions can be made by using the post-fault recovery voltage magnitude measurements as inputs. The accuracy and robustness of the transient stability prediction algorithm with the voltage magnitude measurements was extensively tested under both balanced and unbalanced fault conditions, as well as under different operating conditions, presence of measurement errors, voltage sensitive loads, and changes in the network topology. During the various tests carried out using the New England 39-bus test system, the proposed algorithm could always predict when the power system is approaching a transient instability with over 95% success rate.Index Terms-Nonlinear classifiers, phasor measurement units, support vector machines, transient stability, wide area protection and control.

Using Trajectory Clusters to Define the Most Relevant Features for Transient Stability Prediction Based on Machine Learning Method

“…The inputs to the intelligent classifier were directly sampled values of these variables. Reference [18] selected all the perturbed generator voltage amplitudes as the initial inputs, so the number of inputs increased exponentially with the number of generators, which caused difficulties with online applications. However, in practical application, regardless of the type of electrical variables used as the input, if the measurements of some generators are missing, the input data channel is incomplete, and the mapping output may have large deviation from the actual result.…”

Integrated volt/Var control in modern distribution power systems based on support vector machines

“…The problem of choosing an architecture for a neural network is replaced here by the problem of choosing a suitable kernel for the SVM, as detailed in the following subsections [22][23][24]. In fact, this aspect has allowed the development of somewhat fast training techniques, even with a large number of input variables and big training sets, which is particularly suitable for large power system analysis [25][26][27][28].…”

“…where N is the number of observations, x i ∈ ℜ D are the input data vectors, D is the number of features (e.g., load active and reactive power) of the input data vector and y i ∈ {À1, 1} are the assigned class labels (e.g., capacitor bank switched on or off). The decision function is given by [22][23][24][25]:…”

“…SVM data mining technique has been receiving increasing attention in areas ranging from its original application in pattern recognition to the extended application of regression estimation [22][23][24][25][26][27].Initially, from a previous off-line study, optimal capacitors and voltage regulator settings are found for a particular load profile through any optimization algorithm (or by trial and error) to minimize the energy loss of the whole distribution circuit. Then, each combination of active and reactive power load profile and/or voltage and the resulting optimal control settings are provided as input and output, respectively, for the training of the SVM scheme.…”

Prediction Model of the Power System Frequency Using a Cross-Entropy Ensemble Algorithm

“…AC/DC transmission line failure may cause transmission power to fluctuate greatly, so that the problem of frequency stability is exacerbated [1]. One measure to reduce the risk of a frequency problem occurring is to perform online frequency prediction to determine whether the system is operating under an urgent condition subject to potential power disturbance [2]. Typically, the dynamic process of power system frequency involves multiple time scales ranging from milliseconds to hours.…”