Intelligent techniques approach to power systems identification and control

Identification and control of power systems is one of the major problems of interest in the power system area. Conventional methods of power systems identification and control are very unattractive because they are too cumbersome for on-line applications, based on linear models of power systems, and not easy to implement. The applications of intelligent techniques to power systems identification and control are scrutinized in this dissertation.;A hybrid neuro-fuzzy power system stabilizer using fuzzy basis function network is proposed. The proposed stabilizer combines the different strengths of neural networks and fuzzy logic systems and overcome each other's weaknesses. Unlike the conventional PSS, the proposed stabilizer incorporates the linguistic and numerical information in a uniform fashion.;Incorporating genetic algorithms (GA) into the design of PSS is also proposed. The suggested approach uses GA to search for the optimal settings of PSS parameters. One of the most important features of the proposed approach is the fact that minimal knowledge of the system is required. In addition, an explicit linearized mathematical model of the system is not needed to design the proposed stabilizer.;An approach to integrate the use of GA and rule-based systems to design a genetic rule-based PSS is proposed. The proposed technique is also applied to integrate the use of GA and fuzzy logic systems in order to design a genetic-based fuzzy logic PSS. The proposed approach incorporates GA to search for optimal settings of rule-based and fuzzy logic PSSs parameters and efficiently overcomes the difficulties in design of these stabilizers.;Radial basis function networks (RBFNs) are proposed for off-line as well as on-line identification of synchronous generators. The proposed algorithms are able to capture the nonlinear dynamics of synchronous generators and produce parsimonious models with simple structures. On the control side, a strategy using RBFN to adaptively tune power system stabilizers (PSSs) parameters on-line based on real-time measurements of system operating conditions is proposed.;The proposed identification and control schemes introduced in this dissertation have been tested on several power systems with different complexities and under different disturbances and loading conditions. The results obtained by the proposed schemes are compared with those reported in the literature.;The major features of the proposed schemes are: (1) Easy to tune because of their decentralized nature; (2) Easy to set up and implement using microcomputer; (3) Linguistic and numerical information can be easily incorporated; (4) Far less information than other design techniques is required; (5) Cooperatively work with the existing conventional schemes; (6) Efficiently combine strengths of different intelligent techniques; (7) Properly work over a wide range of operating conditions.