[attachment=13277]
Abstract-
This paper describes the concept andapplications of artificial intelligence techniques for solvingproblems in electrical power systems. It first reports areasin power systems that artificial intelligence has been appliedto. It then summarizes the artificial intelligence techniqueswhich have been employed and makes suggestions for theimprovement of existing artificial intelligence tools. Futurethemes for further development in artificial in powersystems are proposed.
I. INTRODUCTION
Modern power systems are required to generate and supplyhigh quality electric energy to customers. To achieve thisrequirement, computers have been applied to power systemplanning, monitoring and control. Power system applicationprograms for analyzing system behaviors are stored incomputers. In the planning stage of a power system, systemanalysis programs are executed repeatedly. Engineers adjustand modify the input data to these programs according totheir experience and knowledge about the system untilsatisfactory plans are determined. However, among theprograms so far developed for power system,Analysis andplanning are based on mathematical models and areimplemented using languages which are suitable fornumerical computation only.Computer based energy management systems are now widelyused in energy control centers. Power system analysisprograms and other application programs are employed inEnergy Management Systems for the purposes ofinvestigating and predicting the behavior of power systemsunder steady-state operations. The energy managementsystem (EMS) is the center of a control system organized in ahierarchical structure utilizing remote terminal units,communication links, and various levels of computerprocessing systems. The function of the EMS is to ensure thesecure and economic operation of the power system as wellas to facilitate the minute-by-minute tasks carried out by theoperations personnel. While these programs are powerfultools, their ability to assist operation engineers to makeefficient decisions is very limited when unplanned orunexpected modes of system operation occur. The abnormalmodes of system operation may be caused by network faults,active and reactive power imbalances, or frequencydeviations. An unplanned Operation may lead to a completesystem blackout. Under these emergency situations, powersystems are restored back to the normal state according todecisions made by experienced operation engineers.For efficient diagnosis of network faults, determination ofoperational strategies for network restoration, and balancingactive and reactive powers, there is clearly a need to developnew computer techniques and methods to build programswhere the precious knowledge of experienced operationengineers can be accounted for in addition to the conventionalpower system application programs. There is also a need todevelop fast and efficient methods for the prediction ofabnormal system behavior. Artificial intelligence (AI) hasprovided techniques for encoding and reasoning withdeclarative knowledge. It provides conventional computingtechniques and methods for solving problems of power systemplanning, operation and control. This paper first reports areasin power systems that artificial intelligence has been applied to.It then summarizes the artificial intelligence techniques whichhave been employed and makes suggestions for theimprovement of existing artificial intelligence tools.
I. ARTIFICIAL INTELLIGENCE TECHNIQUESEMPLOYED
The research in artificial intelligence has developed manytechniques and methodologies which are useful for solvingcomplicated power system problems. These include knowledgeof representation methods, search strategies, automatedreasoning techniques, expert system or knowledge-basedsystem methodologies, general problem solving approach,blackboard architecture and computer languages for symbolicand list processing. The artificial intelligence techniques andthe expert system approach are some new tools for power engineers
III. EXPERT SYSTEMS
An expert system is a software paradigm where knowledgeconcerning a complex problem. It is encoded into a computerprogram. The framework of expert systems is designed toenable easy encoding of knowledge and easy checkout of theexpert system’s performance. A general architecture for expertsystems is shown in Fig. 1. Four major software elementscomprise an expert system: the knowledge base, an inferenceengine, building and checkout utilities, and the user interface.Expert systems also provide the ability to explain the reasoningused (e.g., to trace the rules used in a rule-based system) whichis important in checking it out. Depending on therepresentation scheme, an AI program becomes either rulebased,frame-based, or logic-based.Fig: Architecture of Expert SystemsA. Rule-Based System:The rule-based system has two kinds of memory: short-term(or working memory) and long-term. The short-term memory(STM) contains factual knowledge, to be modified as theComputation proceeds. The long-term memory (LTM)contains the production rules themselves. The inferenceengines of the rule-based system test the premise-part bymatching it against the factual knowledge in the STM(matching cycle). If it succeeds, the action-part of the rule isexecuted resulting in some changes to the STM (firing cycle).The engine then goes back to the matching cycle. There maybe more than one rule which succeeds in matching and theinference engine then invokes a conflict resolutionmechanism to decide which rule shall be used.The rule-based method was applied to the areas of faultdiagnosis and control of nuclear power plants.B. Frame-Based System:In the rule-based system, factual knowledge is stored in theSTM without regard to relationships between differentobjects. However, a relation between the objects of manyproblems and a frame-based knowledge representation allowsthe user to set up and make use of these relationships.For example, consider the objects of a substation such asbreakers, switches, buses, transformers, and transmissionlines. Several objects comprise a substation, and a set ofsubstations becomes an area. Depending on the status ofindividual breakers and switches, buses may be split or deenergized.Transformers and lines may be connected, openended,or de-energized depending on the status of theterminating bus sections, etc.C. Logic-Based System:The frameworks we have dealt with so far are appropriate torepresent procedural knowledge such as: what to do whencertain conditions are met. A different way to representknowledge requires one to specify “what” instead of “how”. Alogic-based system provides us with such means. Prolog is aprogramming language to represent a “what”-type knowledge.Logic-based systems have an advantage when specifyingsystem requirements, but they have a disadvantage inspecifying procedure-oriented knowledge. Systems developedbased on logic and logic programming has demonstrated thatthese techniques are suitable for building expert systems andartificial intelligence systems for solving power systemcombinatorial problems.
IV. SUGGESTIONS FOR IMPROVING EXISTING AITOOLS
As power systems have complex structures and complicateddecision problems, artificial intelligence tools and expertsystems shells are needed to meet the various requirements inrepresenting power system components and structures and tocontrol the inference mechanisms. For development work,existing tools can offer convenient and helpful softwaredevelopment environments to power engineers for thedevelopment of artificial intelligence systems and expertsystems. However, these tools are not specifically developedfor use in power systems especially in the areas of powersystem monitoring, control and operation. More advancedsoftware tools are required if artificial intelligence technologyis to be adopted widely in real-life power systems. These toolsmay be built on top of existing tools and they should be fastand efficient for on-line power system operation and control.The tools should also have the ability toa) Integrate easily with existing energy management systems.(b) Generate and modify power network configurations anddisplay them graphically.© Accommodate various methods of knowledgerepresentation.(d) Create knowledge bases for different power systemanalysis and control applications.(e) Link efficiently with numerical analysis programs.(f) Incorporate and control different inference mechanismsincluding those which have the ability to reason about timedependent events and deal with uncertainty.(8) Provide adequate explanation facilities.
V. POSSIBLE APPLICATIONS OF AI TO POWERSYSTEM OPERATIONS
A. Alarm Processing: The alarm processing problem is andiagnosis problem. When a serious disruption occurs on thepower system, operators can be overloaded with alarmmessages. Because many of the alarm messages are redundantor present information related to the same event the operatorsmay have difficulty in understanding precisely what hashappened. The use of AI to intercept alarm messages andpresent a concise diagnosis is now under active developmentin several organizations.
B. Switching Operations: Statistics show that about 40percent of the tasks at a power system control center arerelated to operations on circuit breakers and line switches.Therefore, the automation of these tasks should benefitsystem operators. One potential application is the automaticgeneration of switching sequences. Some work has been doneon verification of the switching sequences. Anotherapplication is the identification and isolation of faulted line
C. Restoration Control: A large-scale blackout may happenon a power system, although quite infrequently. The fact thatblackouts happen infrequently makes the operator’s job thatmuch harder because of the limited exposure to solving theproblem of restoring the system. As a result, most controlcenters have restoration plans and attempt to train operatorsin restoration using training simulators. However, the numberof possible ways to restore a power system is very large andcan change depending on the state of critical components atthe time the blackout occurs. To this end, a system whichsupports operators by giving them timely guidance and providesthem with a tool for short term operations planning is quitedesirable.