Artificial intelligence application to security control in power systems

摘要

In power system operation, steady state security control is employed to provideudcontinuous supply to customers and avoid damage to power system plant. Theudsteady state security control includes detection and alleviation of transmissionudequipment overloads and bus voltage violations.udThe three main functions of the security control, ie. security monitoring,udcontingency analysis and control action analysis, can be performed by anudexperienced system operator with the use of conventional optimisation methods.udHowever, such methods require large computation time and rely on mathematicaludmodels and sophisticated programming techniques. They can only cover theudanalytical part of solutions, leaving the burdensome task of making numerousudjudgements to the system operator. These methods may not be used for real-timeudcontrol of large power systems; in particular, under emergency and abnormaludoperating conditions when significant human expertise is required but due toudemotional stress is not readily available. Thus, there is a need for new methodsudand tools such as decision-support systems, to improve the computational speedudand assist operators in making prompt and correct decisions on control actionsudunder emergency and abnormal conditions. During the last decade the computational approach to artificial intelligence (AI)udhas undergone a significant evolution. Results obtained from the internationaludsurveys on AI applications in power systems show that an interest in applicationsudof Al to power system problems is growing strongly. Al is a promisingudtechnology that will be able to fill the gap between human capabilities anduddifficulties involved in the daily operation and planning of modern power systems.udThe work presented in this thesis addresses the aspects of Al applications to steadyudstate security control in power systems. This research focuses on two major issues. Firstly, analysis of the methods employed in steady state security controludand identification of potential Al applications where the existing methods areuddeficient. The second task is the development of decision-support systems toudincorporate operator knowledge, provide an interactive-mode interface to users,udand improve the computational speed. Three decision-support systems have beenuddeveloped:ud• an intelligent system for determination of short-time thermal ratings andudpermissible overload duration of transmission lines using the rule-basedudexpert system and artificial neural network.ud• a prototype expert system for transmission line overload alleviation usinguddatabase, rule-based, and sensitivity tree approaches.ud• a prototype expert system for voltage control and reactive powerudcompensation using rule-based and sensitivity tree approaches.udThe following methods have been proposed and implemented in the developmentudof the above decision-support systems: ud• Determination of permissible overload duration of transmission lines basedudon the short-time thermal rating.ud• Estimation of instantaneous solar radiation employed in the determinationudof thermal rating of transmission lines using artificial neural network andudregression techniques.ud• Estimation of the distance to a voltage collapse using the stability marginudanalysis.ud• Automatic allocation of static and dynamic reactive power compensationudto improve solution of voltage security and stability control.ud• Reduction of the computation time for voltage and reactive power controludusing the "three-tier" network equivalencing technique.udThe decision-support systems have been successfully tested on several powerudsystems, such as the IEEE 30-bus, AEP 57-bus, and real 293-bus power system ofudthe Hydro-Electric Commission of Tasmania. Results obtained show that the decision-support systems provide fast and correct solutions with advice on controludactions expressed in the natural language form. Therefore, the intelligent systemsuddeveloped can be effectively applied to assist operators in the detection andudalleviation of line overload and bus voltage violation problems in power systems.udNineteen refereed technical papers have been published including three inudinternational scientific journals. The research results have been applied to currentudpractice in the Hydro-Electric Commission of Tasmania for planning andudoperation studies.