On the Calculation of Bounds of Probability of System Performance Using Evidence Theory

To evaluate the system performance with limited experimental data and uncertainty information, a method to calculate the bounds of probability of system performance is proposed, in which evidence theory is used. CBF and CPF formulations are deduced as lower and upper bounds of probability of system performance respectively, and the estimations of them using Monte Carlo method are employed by sampling the random intervals of uncertainty parameters. With this method, a less structured representation of uncertainty is possible than is the case with probability theory. The simulation results indicate the proposed algorithm is correct and effective, and can provide a basis for design, performance evaluation, and reliability evaluation of system.