![]() | Only 14 pages are availabe for public view |
Abstract Research work in the area of power system stability is an important task for electrical power engineers. The power system stability problem may be defined as the main problem that accompanied the growth, development, and complex interconnection of power system. The solution ~~ of the problem has been tried out in several directions. Static excitation systems, fast acting automatic voltage regulators, power system stabilizers and even fast controllers or the governor system represented on direction related to the unit level. On the station or area level the local control room with facility on load sharing and load shedding acted as a second direction. The third and the most important is the dispatch among areas on bases of economics or any other consideration. Although modem power systems are complex and may be composed of a large number of generators, the two machine system still represent a challenging problem for the investigation of this system stability. A twomachine system can in conclusion represent a full and large power system such as Egypt’s system. This system can be looked as two-machine system with a long transmission line connected in-between. The thermal power plants in the north and the hydro-electric plants in the south lend them as the two machines with 500 kv line as connectors. The present work investigates the stability problem with two main aims. The first goal is obtaining a fast approximate system of monitoring system transient stability based on the classic equal area criterion. The system parameters in this case required for computation are minimum and so is the computational time. The results of this part are used as an early alarm data telling operators whether the system will be inherently stable or may be driven into a critical stability problem. There are different methods have been suggested to improve the system transient stability such as forced excitation, fast switching, increased inertia and fast valving methods. The second goal is study the system dynamic stability using different types of the power system stabilizers. The conventional power system stabilizer is designed only for a specific operating point. The artificial neural network (ANN) power system stabilizer overcomes this difficulty and successfully used in many other applications.This thesis presents an efficient on-line ANN-based power system stabilizer of the power system. The on-line training technique is used to update the weights and biases of the ANN using the on-line back propagation algorithm. The ANN control technique is used to enhance the power system to avoid the system oscillation when it is subjected to different disturbances. Meanwhile the ANN-PSS is compared with the conventional PSS to demonstrate the advantages of the proposed ANN. The effect of changing the learning rate, the slopes of the linear activation function of the output layer and the number of the neurons in the hidden layer on the dynamic response of the single machine to infinite bus power system are demonstrated. A novel configurations of the ANN driven by a variable learning rate and another configuration of the ANN driven by a variable slope of the linear activation function of the output layer are introduced in order to obtain the best dynamic performance of the single machine to infinite bus disturbed system. The two-machine system under study consists of two synchronous generators each equipped with AVR and PSS. The system is SUbjected to different disturbances on each machine termin al, also the PSS of each machine may be conventional or ANN-based. The proposed ANN-PSS of the disturbed machine takes larger control action than the ANN-PSS of the second machine. The effect of changing the learning rate and the slopes of the linear activation function of the output layer in the ANN-PSSs of the two machines are demonstrated. |