A New Fast and Efficient Artificial Neural Network Based State Estimator Incorporated into a Linear Optimal Regulator for Power System Control Enhancement

作者：Jazaeri Mostafa^*; Nasrabadi Masoud Taji

来源：Electric Power Components and Systems, 2015, 43(6): 644-655.

DOI：10.1080/15325008.2014.997326

• linear optimal regulator
• extended Kalman filter
• feed-forward neural network
• state estimation

Abstract-Fast and accurate state estimation has a crucial role in practical implementation of state feedback-based controllers. But most of the states are not accessible or economically feasible to measure and must be estimated. Although such controllers are normally designed offline, their dynamic performance is evaluated online. Hence, having high speed with acceptable accuracy is an essential feature for estimators. In this article, a state estimator based on an artificial neural network incorporated into a linear optimal regulator is introduced. First, an extended Kalman filter is designed, then an estimator based on a proposed feed-forward neural network structure is elaborated after much effort on promising neural network structures. Different neural networks are trained using the data collected from the extended Kalman filter, and a qualified and shapeable alternative for the extended Kalman filter in reducing its drawbacks is obtained. The optimum structure is identified when minimum state estimation error is achieved. Significant speed and sufficient accuracy are the main advantages of the proposed structure to be used online for state estimation. Dynamic performance of the system equipped with the linear optimal regulator plus different estimators is examined and compared to a single-machine/infinite-bus power system in MATLAB (The MathWorks, Natick, Massachusetts, USA).

• 出版日期2015-4-3