Assessment of the annual frequency and duration indices in composite system reliability using genetic algorithms

A new approach for the assessment of annual frequency and duration indices of composite system is presented. Genetic algorithms (GA) are used as a state sampling tool for the composite generation and transmission system. The GA binary chromosomes are used to represent composite system states. The system hourly load for the year is represented as a multistate component using k-means clustering technique. Transition rates between the load states are calculated. The conditional probability based method is used to calculate the frequency of sampled failure states using different component transition rates. The GA samples network failure states with the system load assigned its maximum state value. Failure states are then reevaluated with lower load states until a success state is obtained or all load states have been evaluated. Through its fitness function GA is able to trace intelligently the network failure states. Case studies on a sample test system are presented. Results are compared with those obtained by nonsequential Monte Carlo simulation. These results are analyzed to indicate the advantages of the proposed method over other conventional methods.