When a permanent fault occurs in distribution systems, the fault location should be detected, isolated and the un-faulted but out of service areas have to be restored. The outage areas can be minimized by the switching operation based on the system load characteristics. By integrating the Outage Management Information System (OMIS), the connectivity of customers and feeder/transformer, the Customer Information System (CIS), the Automated mapping /Facility Management (AM/FM) with the customer load patterns, the hourly load demand and the service priority index of each distribution feeder and each service zone are calculated. By this way, the service restoration of the most power demand and customers can be obtained for the fault contingency of distribution system. To enhance the effectiveness of switching operation for fault contingency of distribution system, the Expert System with CLIPS has been developed by considering the operation rules in the application software program. A underground distribution system with 26 feeders in Kaohsiung District of Taiwan Power Company has been selected for computer simulation to solve the proper switching operation by taking into account the service priority of customers. It has been verified that the proposed methodology can restore the customer power service effectively by Expert System with distribution operation rules.

Title from PDF of title page. / Document formatted into pages; contains 57 pages. / Thesis (M.S.C.S.)--University of South Florida, 2003. / Includes bibliographical references. / Text (Electronic thesis) in PDF format. / ABSTRACT: A clinical trial is defined as a study conducted on a group of patients to determine the effect of a treatment. Assignment of patients to clinical trials is a data and labor intensive task. Usually, medical personnel manually check the eligibility of a patient for a clinical trial based on the patient's medical history and current medical condition. According to studies, most clinical trials are under-enrolled which negatively affects their effectiveness. We have developed web-based agents that can test the eligibility of patients for many clinical trials at once. We have tested various heuristics for optimizing cost and data entry needed in assigning patients to clinical trials. Testing eligibility of a patient for many clinical trials is only feasible if it is cost and data entry efficient. Agents with different heuristics were then tested on data from current breast cancer patients at the Moffitt Cancer Center. / ABSTRACT: Results with different heuristics are compared with each other and with that of the clinicians. It is shown that cost savings are possible in clinical trial assignment. Also, less data entry is needed when probabilistic agents are used to reorder questions. / System requirements: World Wide Web browser and PDF reader. / Mode of access: World Wide Web.

The availability of new treatments for a disease depends on the success of clinical trials. In order for a clinical trial to be successful and approved, medical researchers must first recruit patients with a specific set of conditions in order to test the effectiveness of the proposed treatment. In the past, the accrual process was tedious and time-consuming. Since accruals rely heavily on the ability of physicians and their staff to be familiar with the protocol eligibility criteria, candidates tend to be missed. This can result and has resulted in unsuccessful trials.A recent project at the University of South Florida aimed to assist research physicians at H. Lee Moffitt Cancer Center & Research Institute, Tampa, Florida, with a screening process by utilizing a web-based expert system, Moffitt Expedited Accrual Network System (MEANS). This system allows physicians to determine the eligibility of a patient for several clinical trials simultaneously.We have implemented this web-based expert system at the H. Lee Moffitt Cancer Center & Research Gastroenterology (GI) Clinic. Based on our findings and staff feedback, the system has undergone many optimizations. We used data mining techniques to analyze the medical data of current gastrointestinal patients. The use of the Apriori algorithm allowed us to discover new rules (implications) in the patient data. All of the discovered implications were checked for medical validity by a physician, and those that were determined to be valid were entered into the expert system. Additional analysis of the data allowed us to streamline the system and decrease the number of mouse clicks required for screening. We also used a probability-based method to reorder the questions, which decreased the amount of data entry required to determine a patient's ineligibility.

Στα πλαίσια αυτής της διπλωματικής εργασίας, δημιουργήθηκε ένα τραπεζικό έμπειρο σύστημα (ΤΕΣ) για την αξιολόγηση πελατών και εύρεση κατάλληλου δανείου, ανάλογα με τις απαιτήσεις του πελάτη. Τα δάνεια που διαπραγματεύεται το ΤΕΣ δεν αφορούν μόνο μία, αλλά περισσότερες τράπεζες, ώστε το σύστημα να μπορεί να χαρακτηριστεί διατραπεζικό. Για την επίλυση του προβλήματος και την υλοποίηση του (ΤΕΣ), χρησιμοποιήθηκαν 2 τρόποι. Κατ’ αρχήν υλοποιήθηκε στο Fuzzy Clip, ένα εργαλείο ανάπτυξης ευφυών συστημάτων, που χρησιμοποιεί ασαφής κανόνες για την αναπαράσταση γνώσης. Ηχρήση ασάφειας ήταν ταιριαστή , καθώς αρκετές απο τι παραμέτρους μπορούσαν να θεωρηθούν ώς τετοιες. Στο δεύτερο τρόπο χρησιμοποιήθηκαν εμπειρικά δεδομένα, δηλ. συγκεκριμένες περιπτώσεις δανειοληπτών, για την δημιουργία του ΤΕΣ. Για το σκοπό αυτό χρησιμοποιήθηκε το εργαλείο HYMES (Hybrid Modular Expert System), ένα εργαλείο που χρησιμοποιεί υβριδικούς κανόνες για αναπαράσταση γνώσης, των νευροκανόνων. Οι νευροκανόνες συνδυάζουν νευρωνικά δίκτυα και συμβολικούς κανόνες και παράγονται από εμπειρικά δεδομένα. Τέλος έγινε συγκριση των δυο συστημάτων με βάση τρεις μετρικές που χρησημοποιούνται σ’ αυτές τις περιπτώσεις. / In the framework of this senior project, a banking intelligent system (BIS)has been developed which is suitable for the customers evaluation and the finding of the appropriate loan according to the customer’s demands. The loans that the (BIS) has been developed which is suitable for the customers evaluation and the finding of the appropriate loan according to the customer’s demands. The loans that the BIS treats, concern more than one bank so that for the system to be characterized as a multi- banking system. For the solution of the problem and the implementation of the BIS, two methods have been used. Firstly, the Fuzzy Clips was developed, which is a tool that is used to form intelligent systems and used fuzzy rules to represent knowledge. The usage of the indefiniteness was done successfully since many of the parameters used could be considered as fuzzy. In the second method, experiential data have been used in order to create the BIS. This means that data have been used in order to create the BIS. This means that specific real cases of loan receiving were taken in mind. For this purpose, the HYMES (Hybrid Modular Expert System) was used which is a tool that used Hubrid rules to represent knowledge. These rules are known as neurotic rules, they combine neuronal net and symbolic rules and they are derived from experiential data. Finally, a comparison of the above two systems was made based on three metrics that are used in those occasions.

Η παρούσα διατριβή πραγματεύεται την ανάπτυξη μιας καινοτόμου μεθοδολογίας για την βελτιστοποίηση τόσο του σχεδιασμού όσο και της λειτουργίας βιομηχανικών εγκαταστάσεων. Η μεθοδολογία αυτή βασίζεται στη χρήση δυο μεθόδων Τεχνητής Νοημοσύνης (Έμπειρων Συστημάτων και Γενετικών Αλγόριθμων) για τη δημιουργία ενός λογισμικού το οποίο λαμβάνοντας από το χρήστη στοιχεία σχετικά με τα κριτήρια σχεδιασμού (ή ανασχεδιασμού) μιας διεργασίας καθώς και για τις συνθήκες λειτουργίας της θα εξαγάγει τόσο τις βέλτιστες διαθέσιμες τεχνολογίες για τη διεργασία αυτή όσο και τις βέλτιστες συνθήκες λειτουργίες των επιλεχθεισών τεχνολογιών. Προκειμένου να λειτουργήσει η αναπτυχθείσα μεθοδολογία πέρα από τα στοιχεία τα οποία παρέχει ο χρήστης είναι απαραίτητη η ύπαρξη μιας βάσης δεδομένων η οποία θα περιέχει τις διαθέσιμες τεχνολογίες οι οποίες είναι δυνατόν να χρησιμοποιηθούν για τη συγκεκριμένη διεργασία, δηλαδή όλα τα τμήματα εξοπλισμού που μπορούν να χρησιμοποιηθούν για τη διεργασία αυτή. Οι βέλτιστες διαθέσιμες τεχνολογίες, τις οποίες η μεθοδολογία επιλέγει από αυτή τη βάση δεδομένων, είναι τα τμήματα του εξοπλισμού εκείνα τα οποία πληρούν με τον καλύτερο δυνατό τρόπο τα κριτήρια σχεδιασμού της διεργασίας. Για την εύρεση των βέλτιστων τεχνολογιών για μια διεργασία για κάθε διαθέσιμη τεχνολογία πρέπει να λαμβάνονται υπόψη και τα εκτιμώνται: τεχνικά κριτήρια όπως τα όρια λειτουργίας της, η ολοκλήρωσή της στη διεργασίας (δηλαδή η διασύνδεσή της με άλλες τεχνολογίες που χρησιμοποιούνται στην ίδια διεργασία), οικονομικά κριτήρια όπως το κόστος εγκατάστασης, το κόστος λειτουργίας, ή / και το κόστος συντήρησης, περιβαλλοντικά στοιχεία όπως οι εκπομπές ρύπων, παραπροϊόντα. Η απαίτηση για ταυτόχρονη ικανοποίηση όλων αυτών των κριτηρίων καθιστά την εύρεση των βέλτιστων διαθέσιμων τεχνολογιών ένα περίπλοκο πρόβλημα για την επίλυση του οποίου απαιτείται εξειδικευμένη γνώση. Στην αναπτυχθείσα μεθοδολογία η γνώση αυτή καταχωρείται με κατάλληλο τρόπο στο σύστημα και κατά συνέπεια με τη χρήση του είναι δυνατόν ακόμη μη εξειδικευμένα άτομα να βρουν τις βέλτιστες διαθέσιμες τεχνολογίες (και τις βέλτιστες παραμέτρους λειτουργίας του). Η αναπτυχθείσα μεθοδολογία εφαρμόστηκε σε τρία προβλήματα ανασχεδιασμού του συστήματος συμπαραγωγής ενέργειας ενός διυλιστηρίου. Από τα αποτελέσματα που προέκυψαν από τις εφαρμογές αυτές γίνεται φανερό ότι η μεθοδολογία καταλήγει σε βέλτιστες λύσεις του προβλήματος για τις εκάστοτε συνθήκες και ότι ο αλγόριθμος της μεθοδολογίας είναι ιδιαίτερα εύρωστος υπό την έννοια ότι η σύγκλισή του δεν επηρεάζεται από τις τιμές των παραμέτρων που χρησιμοποιούνται για την εκτέλεσή του. / The thesis at hand deals with the development of a novel methodology for the optimisation of both the design and the operation of industrial plants. The methodology is based on two Artificial Intelligence techniques (Expert Systems and Genetic Algorithms) for the development of a software which, given from the user data related to the design criteria of the process, as well as the conditions of operation of the process, it will output both the best available technologies for that process as well as the optimal working conditions of the selected technologies. Apart from the data provided from the user, the methodology requires also a database which should contain the available technologies which can be used for the process at hand. These technologies are the various components of the equipment which can be used for the process. The best available technologies which are selected by the methodology from the database are those components which satisfy in the best possible way the design criteria. For the identification of the best technologies for a process the methodology assess (for each of the available techonologies): technical criteria such range of operation, its integration to the whole process (i.e. its connectivity with other technologies used in the process), financial criteria such as the cost of installation, the cost of operation and or the cost of maintenance, environmental criteria such as emissions of various pollutants, side products etc. The requirement of the simultaneous satisfaction of these criteria makes the identification of the best available technologies a complex problem which requires specific knowledge (expertise) in order to be solved. In the developed methodology this specific knowledge (expertise) is stored in a proper way in the system and therefore it is possible that even not-expert users of the system to identify the best available technnologies (and the best working conditions of the selected technologies). The developed methodology was applied to three problems of re-design of the cogeneration plant of a refinery. From the results obtain from these applications it is evident that the methodology converges to near optimal solutions for the criteria set each time and that the algorithm of the methodology is robust since its convergence is not affected by the value set for the algorithm's parameters during each of the runs.

Η Τεχνολογία της Γνώσης, και πιο συγκεκριμένα τα Έμπειρα Συστήματα (expert systems) ή αλλιώς Συστήματα Γνώσης (knowledge based systems) αποτελούν τους κλάδους της Τεχνητής Νοημοσύνης που αποδεδειγμένα έχουν προσφέρει τα πιο αξιοσημείωτα αποτελέσματα στην πρακτική τους εφαρμογή. Ως έμπειρα συστήματα θεωρούνται προγράμματα τα οποία επιδεικνύουν νοήµονα συµπεριφορά σε συγκεκριµένους τοµείς και διαδικασίες, ανάλογη με εκείνη ενός ανθρώπου εµπειρογνώµονα µε ειδικότητα στον ίδιο τοµέα. Η λειτουργία των έμπειρων συστημάτων βασίζεται τόσο στην κωδικοποίηση της γνώσης και της συλλογιστικής του ανθρώπου-ειδικού σε έναν εξειδικευμένο τομέα όσο και στο χειρισμό αυτής της γνώσης, με κυριότερους στόχους τη διόρθωση βλαβών, την πρόβλεψη, παρακολούθηση και ερμηνεία καταστάσεων, τη διαμόρφωση και τον έλεγχο συστημάτων. Κατά τις τελευταίες δεκαετίες, η σημαντική πρακτική επιτυχία της εφαρμογής έμπειρων συστημάτων σε εξειδικευμένους τομείς όπως η φαρμακευτική διάγνωση (MYCIN), η ανάλυση γεωλογικών δεδομένων για εντοπισμό πετρελαίου (DIPMETER) και μετάλλων (PROSPECTOR), και η διαμόρφωση υπολογιστικών συστημάτων (XCON), έχουν οδηγήσει σε μία έκρηξη ενδιαφέροντος ως προς τη χρήση έμπειρων συστημάτων σε πολύ ευρύτερο φάσμα εφαρμογών. Το αντικείμενο της παρούσας διπλωματικής εργασίας είναι η ανάπτυξη ενός έμπειρου συστήματος που θα συνάγει στον έλεγχο της κατάστασης λειτουργίας της κύριας μηχανής φορτηγών πλοίων τύπου container, με στόχο την πρόληψη ή έγκαιρη διόρθωση βλαβών και εν τέλει την επίτευξη αποδοτικότερης και ασφαλέστερης λειτουργίας της μηχανής, λιγότερων φθορών και μεγαλύτερου χρόνου ζωής. Η ανάγκη εφαρμογής ενός τέτοιου συστήματος, που θα παρέχει αντικειμενικά και συνεπή συμπεράσματα, από τις εταιρείες διαχείρισης των πλοίων γίνεται όλο και μεγαλύτερη. Όχι μόνο γιατί αποτελεί βασικό μέρος του συστήματος διαχείρισης της ποιότητας (quality management system) των παρεχόμενων υπηρεσιών της εταιρείας προς τους πελάτες – ναυλωτές των πλοίων τύπου container, αλλά κυρίως γιατί μπορεί να αποτελέσει βασικό αρωγό στην προσπάθεια της εταιρείας για συμμόρφωση με διεθνή πρότυπα και κανονισμούς περιβαλλοντικής διαχείρισης (environmental management) για την προστασία εκείνων του στοιχείων του περιβάλλοντος στα οποία έχει επιπτώσεις η δραστηριότητα της εταιρείας. Για τη διαδικασία απόκτησης της γνώσης που θα αποτελέσει την εμπειρογνωμοσύνη του συστήματος ή αλλιώς τη βάση γνώσης του συστήματος (knowledge base) θα χρησιμοποιηθεί η μέθοδος των μη-δομημένων και ημιδομημένων συνεντεύξεων με μηχανολόγους μηχανικούς, εξειδικευμένους σε θέματα μηχανών του συγκεκριμένου τύπου. Τα δεδομένα πάνω στα οποία θα εφαρμοστεί η βάση γνώσης του συστήματος (working memory), ώστε να εξαχθούν τα συμπεράσματα για τη λειτουργία της μηχανής προέρχονται από μετρήσεις πάνω στα όργανα της μηχανής ανά τακτά χρονικά διαστήματα καθώς και δεδομένα-αναφοράς που αντιπροσωπεύουν την ιδανική κατάσταση λειτουργίας της μηχανής. Το κέλυφος του συστήματος (expert system shell) θα αποτελέσει η μηχανή JBoss Rules, η οποία στο επίπεδο του μηχανισμού εξαγωγής συμπερασμάτων (inference engine) υλοποιεί μία αντικειμενοστραφή παραλλαγή του αλγορίθμου Rete, ενώ επιπλέον παρέχει σύστημα διαχείρισης της βάσης γνώσης που καθιστά την ανάπτυξη, τροποποίηση και τον έλεγχό της δυνατό ακόμα και από τους ίδιους τους εμπειρογνώμονες. / Knowledge Technology, and more concretely Expert Systems or Knowledge Based Systems, constitute the sectors of Artificial Intelligence that are proved to have succeeded the most remarkable results in their practical application. Expert Systems are considered to be the programs that demonstrate intelligent behaviour in specific areas and processes, proportional to that of a human expert with speciality in same area. The operation of Expert Systems is based so much on the coding of knowledge and reasoning of a human expert specialized in the specific area, as in the handling of this knowledge, with main objectives the correction of damage, the provision, monitor and interpretation of conditions, the configuration and the inspection of systems. At the last decades, the important practical success of application of Expert Systems in specialised areas such as pharmaceutical diagnosis (MYCIN), the analysis of geological data on location of oil (DIPMETER) and metals (PROSPECTOR), and the configuration of computer systems (XCON), have led to an explosion of interest as for the use of Expert Systems to much wider spectrum of applications. The object of the present thesis is the implementation of an Expert System that will assist in the inspection of the condition of operation of Container Ships Main Engines, aiming at the prevention or on time correction of damage, and finally at achieving more efficient and secure operation of the main engine, less deteriorations and effectively greater life span. The need of such a system, which will provide objective and reliable conclusions, being applied by the shipping management companies becomes even greater. Not only because it constitutes basic part of the quality management system of the provided by the company services to the customers - freighters of the container ships, but mainly because it can constitute basic helper in the company effort to conform with international models and regulations of environmental management on the protection of those aspects of the environment that are affected by the company’s activity. For the process of acquiring the knowledge that will constitute the expertise of the system or differently the knowledge base of the system we will use the method of non-structured and semi-structured interviews with mechanical engineers, specialised on issues regarding the specific engines. The data, on which the working memory of the system will be applied, so that the conclusions on the function of the engine will be exported, emanate from measurements on the different parts of the engine per regular time intervals as well as benchmark data that represent the ideal condition of the engine. The expert system shell will be constituted by the JBoss Rules Engine, which in the level of the inference engine implements an object-oriented variant of the Rete algorithm, while moreover it provides an administration system for the knowledge base that makes its evolution, modification and inspection possible even for the experts themselves.

This thesis describes the analysis of over 1300 spelling and typing errors. It introduces and describes many empirical rules which these errors obey and shows that a vast majority of errors are variations on some 3000 basic forms. It also describes and tests an intelligent, knowledge based spelling error correction algorithm based on the above work. Using the Shorter Oxford English dictionary it correctly identifies over 90% of typical spelling errors and over 80% of all spelling errors, where the correct word is in the dictionary. The methodology used is as follows: An error form is compared with each word in that small portion of the dictionary likely to contain the intended word, but examination of improbable words is rapidly abandoned using heuristic rules. Any differences between the dictionary word and the error form are compared with the basic forms. Any dictionary word which differs from the error form only by one or two basic forms is transferred to a separate list. The program then acts as an expert system where each of the basic forms is a production or rule with a subjective Bayesian probability. A choice is made from the list by calculating the Bayesian probability for each word in the separate list. An interactive spelling error corrector using the concepts and methods developed here is operating on the Bradford University Cyber 170/720 Computer, and was used to correct this thesis. The corrector also runs on VAX and Prime computers.

The bioreactor landfill is an emerging concept for solid waste management that has gained significant attention in the last decade. This technology employs specific operational practices to enhance the microbial decomposition processes in landfills. However, the unsupervised management and lack of operational guidelines for the bioreactor landfill, specifically leachate manipulation and recirculation processes, usually results in less than optimal system performance. Therefore, these limitations have led to the development of SMART (Sensor-based Monitoring and Remote-control Technology), an expert control system that utilizes real-time monitoring of key system parameters in the management of bioreactor landfills. SMART replaces conventional open-loop control with a feedback control system that aids the human operator in making decisions and managing complex control issues. The target from this control system is to provide optimum conditions for the biodegradation of the refuse, and also, to enhance the performance of the bioreactor in terms of biogas generation. SMART includes multiple cascading logic controllers and mathematical calculations through which the quantity and quality of the recirculated solution are determined. The expert system computes the required quantities of leachate, buffer, supplemental water, and nutritional amendments in order to provide the bioreactor landfill microbial consortia with their optimum growth requirements. Soft computational methods, particularly fuzzy logic, were incorporated in the logic controllers of SMART so as to accommodate the uncertainty, complexity, and nonlinearity of the bioreactor landfill processes. Fuzzy logic was used to solve complex operational issues in the control program of SMART including: (1) identify the current operational phase of the bioreactor landfill based on quantifiable parameters of the leachate generated and biogas produced, (2) evaluate the toxicological status of the leachate based on certain parameters that directly contribute to or indirectly indicates bacterial inhibition, and (3) predict biogas generation rates based on the operational phase, leachate recirculation, and sludge addition. The later fuzzy logic model was upgraded to a hybrid model that employed the learning algorithm of artificial neural networks to optimize the model parameters. SMART was applied to a pilot-scale bioreactor landfill prototype that incorporated the hardware components (sensors, communication devices, and control elements) and the software components (user interface and control program) of the system. During a one-year monitoring period, the feasibility and effectiveness of the SMART system were evaluated in terms of multiple leachate, biogas, and waste parameters. In addition, leachate heating was evaluated as a potential temperature control tool in bioreactor landfills. The pilot-scale implementation of SMART demonstrated the applicability of the system. SMART led to a significant improvement in the overall performance of the BL in terms of methane production and leachate stabilization. Temperature control via recirculation of heated leachate achieved high degradation rates of organic matter and improved the methanogenic activity.

This project details the development of a new framework known as the Coordinated Knowledge Management method to enable complete task integration of all low and midlevel tasks for process industries. The framework overcomes past problems of task integration, which made it impossible to have a fully integrated system and with integration being limited to data acquisition, regulatory control and occasionally supervisory control. The main component of the project includes the use of hierarchically structured timed place Petri nets, which have not previously been used for integrating tasks in intelligent process operations management. Tasks which have been integrated include all low-level tasks such as data acquisition, regulatory control and data reconciliation, and all mid-level tasks including supervisory control and most significantly the integration of process monitoring fault detection and diagnosis. The Coordinated Knowledge Management method makes use of hierarchical timed place Petri nets to (i) coordinate tasks, (ii) monitor the system, (iii) activate tasks, (iv) send requests for data updates and (iv) receive notice when tasks are complete. Visualization of the state of the system is achieved through the moving tokens in the Petri net. The integration Petri nets are generic enough to be applied to any plant for integration using existing modules thus allowing the integration of different tasks, which use different problem solving methodologies. Integrating tasks into an intelligent architecture has been difficult to achieve in the past since the developed framework must be able to take into account information flow and timing in a continuously changing environment. In this thesis Petri nets have been applied to continuous process operations rather than to batch processes as in the past. In a continuous process, raw materials are fed and products are delivered continuously at known flow-rates and the plant is generally operated at steady state (Gu and Bahri, 2002). However, even in a continuous process, data is received from the distributed control system (DCS) at discrete time intervals. By transforming this data into process events, a Petri net can be used for overseeing process operations. The use of hierarchical Petri nets as the coordination mechanism introduces inherent hierarchy without the rigidity of previous methods. Petri nets are used to model the conditions and events occurring within the system and modules. This enables the development of a self-monitoring system, which takes into account information flow and timing in a continuously changing environment. Another major obstacle to integration of tasks in the past has been the presence of faults in the process. The project included the integration of fault detection and diagnosis a component not integrated into current systems but which is necessary to prevent abnormal plant operation. A novel two-step supervisory fault detection and diagnosis framework was developed and tested for the detection and diagnosis of faults in large-scale systems, using condition-event nets for fault detection and Radial Basis Function neural networks for fault diagnosis. This fault detection and diagnosis methodology detects and diagnoses faults in the early stages of fault occurrence, before fault symptoms propagate throughout the plant. The Coordinated Knowledge Management method and the newly developed fault diagnosis module were developed in G21 and applied and tested on the Separation and Heating sections of the Pilot plant for the Bayer process at the School of Engineering Science, Murdoch University. Testing indicated that the use of an intelligent system comprising of Petri nets for integration of tasks results in improved plant performance and makes the plant easier to monitor increasing profits. The fault detection and diagnosis module was found to be useful in detecting faults very early on and diagnosing the exact location of faults, which would otherwise prove to be difficult to detect. This would also increase plant safety, reduce wastage and improve environmental considerations of the plant. References Gensym Corporation (1999). G2 Reference Manual (Version 5.0), Cambridge, MA. Gu, T. and P. Bahri (2002). A Survey of Petri Net Applications in Batch Processes, Computers in Industry, 47 (1), p. 99 – 111.

[pt] O presente trabalho investiga a aplicação de um sistema híbrido baseado em Redes Neurais Artificiais (RNAs) e Sistemas Especialistas (SE) na detecção e diagnóstico de falhas em redes elétricas. A pesquisa consistiu de três partes principais:o estudo do problema da detecção e diagnóstico de falhas em redes elétricas; a modelagem e o desenvolvimento do sistema híbrido (RNAs e SE); e o estudo de casos. Na fase de estudo do problema, investigou-se a importância da detecção e do diagnóstico de falhas em redes elétricas, concentrando-se em sistemas elétricos dotados de grande quantidade de alarmes. Tais alarmes podem ocorrer simultaneamente. Conseqüentemente, os operadores de sistemas elétricos encontram dificuldades na identificação da falha e na tomada de decisão quanto à ação corretiva a ser adotada, cometendo, eventualmente, erros de diagnóstico. A investigação do problema também envolveu entrevistas com especialistas da área, visando não só de absorver conhecimento específico sobre o problema, como também delinear a melhor solução para resolvê-lo. A modelagem do sistema híbrido envolveu duas partes:a detecção das falhas, executada por um conjunto de RNAs; e o diagnóstico das falhas detectadas , realizado por um SE. Na detecção das falhas, um conjunto de quatro RNAs, cada uma especializada em um componente do sistema elétrico (gerador, transformador, barra e linha), tem a função de mapear grupos de alarmes acionados em falhas específicas. Trata-se, portanto, de um problema típico de classificação de padrões, onde cada rede neural é treinada usando-se o algoritmo de retropropagação. Os padrões de treinamento, fornecido por especialistas da área, consistem de combinações de 149 alarmes, para um total de 198 ocorrências (184 falhas simples mais 14 situações de operação normal). Após treinadas, as RNAs são testadas com amostras que refletem o estado dos alarmes em um certo período de funcionamento do sistema elétrico. As saídas das RNAs indicam, através de um código, a ocorrência de falhas ou o funcionamento normal do sistema elétrico, nesse período de observação. O SE, responsável pelo diagnóstico, recebe a saída numérica das RNAs, referente às falhas detectadas, e fornece ao operador informações importantes, tais como: quais alarmes foram acionados; quais equipamentos de proteção estão envolvidos na ocorrência; quais os motivos prováveis da ocorrência da falha; e, finalmente, sugere um conjunto de ações corretivas que podem ser tomadas pelo operador para solucionar o problema. Essas informações, não disponíveis diretamente na saída das RNAs, são obtidas através da aplicação de um conjunto específico de especialistas da área. O ambiente de simulação foi desenvolvido em plataforma PC. As RNAs foram implementadas em MatLab Vers: 4.2 e o SE, em Delphi Vers: 2.0. O estudo de casos empregou cerca de 1000 padrões de teste correspondentes ao estado dos 149 alarmes. Estes dados, fornecidos por especialistas do setor elétrico, foram obtidos através de adaptações de situações reais, adequadas ás dimensões do sistema elétrico adotado. Nos testes realizados, o sistema híbrido foi submetido a um conjunto de alarmes, afetados ou não por ruído, respondendo com sugestões quanto às ações corretivas que podem ser tomadas pelo operador. Foram realizados testes de falhas em geradores, tranformadores, barramentos e linhas de transmissão, envolvendo falhas simples e múltiplas no sistema elétrico. Com incidência de até 10% de ruído nos padrões de teste, o índice de acerto de detecção de falha é próxima de 100% e para taxas superiores a 20% o desempenho do sistema híbrido cai gradualmente. Segundo a avaliação de especialistas do setor elétrico, o sistema híbrido apresenta resposta rápida e segura, quando comparado com os processos tradicionais, / [en] This work examines the application of hybrid systems based on Artificial Neural Networks (ANN) and Expert Systems (ES) in detecting and diagnosing faults in Electrical Systems. The research consists of three main parts: the study of cases. In the study of problem, was examined, the importance of detecting and diagnosing faults in Electrical Systems concentrating in Electric Systems equipped with a large quantity of alarms. These alarms may occur simultaneously. Consequently, it is difficult for the electrical systems´operators to identify the faults and decide the corrective action to be adopted, resulting, eventually, in diagnosis erros. The analysis of the problem also involved some interviews with experts in the area, in order to absorb the specific knowledge about the problem, and design the best solution to solve it. The modeling of the Hybrid System involved two parts: the detection of faults, executed by a group of ANNs; and the diagnosis of the detected faults, fulfilled by the ES. In the detection module, a group of four ANNs, each one specialized in an electrical system component (generators, transformers, buses and transmission lines) maps groups of alarms in the specific faults. Therefore, this is a typical pattern classification problem, where each neural network is trained by using the error backpropagation algorithm. The training patterns, produced by experts in the area, consist of the combination of 149 alarms for a total of 184 simple faults and 14 normal operation situations. After training, the ANNs are tested with new samples of alarms, reflecting a certain configuration of the electrical system during the observation period. The ES module, responsible for the diagnosis, receives the ANNs outputs related to the detected faults, and provides to the operator important informations such as: Which alarms were started; the protective equipment involved in the occurrence; the problable reason for the occurrence of the faults; and finally, suggests the corrective action that the operator should perform in order to solve the problem. This information, not available in the ANNs outputs, can be obtained through the application of a set of production rules in a data base, containing the specific knowledge that were extracted from the experts in the area. The simulation environment was developed in a PC plataform. The ANNs, were implemented in MatLab Vers.4.2 and the ES in Delphi. The case studies, applied about 1000 test patterns corresponding to the situation of the 149 alarms. These data, provided by experts of the electrical sector, are adapted from real situations to the dimensions of the Electrical System adopted. In the tests performed, the Hybrid System is submitted to a group of alarms, affected or not by the noise, and reply with suggestions regarding corrective actions that can be adopted by the operator. Various tests were carried out in the generators, transformers, buses, and transmission lines involving simple and multiple faults in the Electrical Power System. With incidence of up to 10% of noise in the test pattern, the performance in detecting fault is near of 100% and for rates superior of 20%, decreases gradually. The evaluation of experts in the electrical sector shows that, the Hybrid System presents a quicker and safer answer, when compared with traditional processes, totally dependent on the human being.