Non-linear LS-SVM with RBF-kernel-based approach for AGC of multi-area energy systems

Support vector machine (SVM) is very efficacious in pattern recognition and function estimation problems. However, its application to design efficient automatic generation control (AGC) scheme has not been investigated till date by the researchers and power engineers worldwide. Hence, in this study an early attempt has been made to present a combined design of optimal control and optimisation problem of least-squares SVM (LS-SVM)-based AGC of multi-area energy systems. The radial basis function (RBF) kernel is used to train the LS-SVM. To validate the efficacy of the approach, the proposed control method is implemented on various 2-area system models such as non-reheat thermal, reheat thermal and hydro-thermal energy systems. The trained LS-SVM RBF-kernel-based AGC is realised with 1% load perturbation in one of the control area and the obtained results are compared with multi-layer perceptron artificial neural network and the conventional integral based controller in order to show the supremacy of the proposed control design. Furthermore, the performance of the AGC design is examined considering the system non-linearities such as governor dead-band and generation rate constraint.