PREDICTION BASED DIAGNOSIS USING ARTIFICIAL INTELLIGENCE METHODS Theses of PhD dissertation

Equipment failures of large and complex safety-critical plants are unavoid-able. The forthcoming fault, failure changes the properties of the system and therefore its operation. In case of faulty operation, the fastest and most accu-rate fault localization is necessary, and with the help of this information the adverse consequences have to be determined. If there is enough information it is possible for the fault to be be identified in time and corrections can be made. Fault Detection and Diagnosis (FDD) is important and lately it has be-come more and more conspicuous in the area of process engineering as well as in other engineering areas. Abnormal Event Management (AEM) [1] has also gained a lot of attention nowadays. The aim of AEM is detection, di-agnosis and correction of errors occurring in a system. Fault Detection and Diagnosis is one of the main parts of Abnormal Event Management. The early identification and diagnosis of errors helps to prevent the spread of ab-normal events and decreases productivity loss while the plant operates in