This chapter is demonstrating a practical design of an intelligent type of controller using higher order neural network (HONN) concepts, for the excitation control of a practical power generating system. This type of controller is suitable for real time operation, and aims to improve the dynamic characteristics of the generating unit by acting properly on its original excitation system. The modeling of the power system under study consists of a synchronous generator connected via a transformer and a transmission line to an infinite bus. For comparison purposes and also for producing useful data in order for the demonstrating neural network controllers to be trained, digital simulations of the above system are performed using fuzzy logic control (FLC) techniques, which are based on previous work. Then, two neural network controllers are designed and applied by adopting the HONN architectures. The first one utilizes a single pi-sigma neural network (PSNN) and the significant advantages over the standard multi layered perceptron (MLP) are discussed. Secondly, an enhanced controller is designed, leading to a ridge polynomial neural network (RPNN) by combining multiple PSNNs if needed. Both controllers used, can be pre-trained rapidly from the corresponding FLC output signal and act as model dynamics capturers. The dynamic performances of the fuzzy logic controller (FLC) along with those of the two demonstrated controllers are presented by comparison using the well known integral square error criterion (ISE). The latter controllers, show excellent convergence properties and accuracy for function approximation. Typical transient responses of the system are shown for comparison in order to demonstrate the effectiveness of the designed controllers. The computer simulation results obtained show clearly that the performance of the developed controllers offers competitive damping effects on the synchronous generator’s oscillations, with respect to the associated ones of the FLC, over a wider range of operating conditions, while their hardware implementation is apparently much easier and the computational time needed for real-time applications is drastically reduced.
TopIntroduction
The problem of electric power system dynamic stability has received growing attention over the last decades. The main reasons for this are the increasing size of generating units and the use of high-speed excitation systems. The effect of the high-speed excitation on dynamic stability is to add negative damping to the system thereby causing oscillations with weak damping. A design of such an excitation system should also be satisfactory for a wide range of operating conditions as well as for fault conditions. Practical methods for nonlinear control include an open-loop inverse model of the nonlinear plant dynamics and the use of feedback loops to cancel the plant nonlinearities. The approximation of a non-linear system with a linearized model yields to the application of adaptive control, where real-time measurements of the plant inputs are used, either to derive explicitly the plant model and design a controller based on this model (indirect adaptive control), or to directly modify the controller output (direct adaptive control). Typical studies concerning applications of modern algebraic and optimal control methods in excitation controller design using linear system models and output feedback have been presented before, (e.g. Papadopoulos, 1986; Papadopoulos, Smith & Tsourlis, 1989; Mao, Malik & Hope, 1990, Karnavas & Pantos, 2008).
The aim of this chapter is to investigate the use of HONNs as a replacement of an existing (designed previously) FLC, applied in the excitation part of a practical synchronous electric machine workbench system. The first type of the polynomial neural network used is called pi-sigma network. This network utilizes product cells as the output units to indirectly incorporate the capabilities of HONNs, while using a fewer number of weights and processing units. The motivation here is to develop a systematic type of controller which maintains the fast learning property of single-layer HONNs, while avoiding the exponential increase in the number of weights and processing units required. The network has a regular structure, exhibits much faster learning, and is amenable to the incremental addition of units to attain a desired level of complexity. If such an incremental addition of units takes place, then a ridge polynomial network (RPNN) is produced. The second controller described here refers to this kind of architecture. Simulation results show excellent convergence properties and accuracy for function approximation. Comparative results using a FLC output training data set are also provided to highlight the learning, and subsequently control, abilities of the proposed PSNN and RPNN controllers.
At first, digital simulations of the above system are performed with FLC controller based on previous works by the author under various disturbance conditions (Karnavas & Papadopoulos, 2000). Next, an effort is made on the design and simulation of the HONN controllers. The new results are compared with those of the FLC. The overall evaluation of the proposed PSNN and RPNN controllers is made through the ISE criterion.