The application of fuzzy logic to the design of on-line monitoring systems

In a transformer monitoring system, such operating parameters as top oil temperature, hot spot temperature, relative corona, hydrogen concentration in transformer oil, and change of hydrogen in oil are monitored and fed to the computer for extensive analysis. The higher the value of a particular parameter, the greater the severity of the threat to the security of the system. The different values of parameters to be monitored are divided into linguistic labels with a membership function associated with each label. Based on system operator´s expert knowledge, rules are written. These rules are expressed using linguistic variables for knowledge representation, because most of the knowledge acquired from the experts is in terms of linguistic labels. There is a weight associated with each rule, which provides relative precedence of one rule over another. Such a system can provide a prompt indication of the system severity. The effectiveness of the proposed fuzzy logic transformer monitoring system is demonstrated by the various practical scenarios