Abstract: To satisfy quantifying evaluation and determine parameters, an evolutionary neural network(ENN) and a multiple objective evolutionary algorithm(MOEA) were presented for infrared thermography nondestructive testing. By means of the measurement of the temperature of specimen with artificial defects, the time against temperature signal was recorded as the network characteristic parameters. The errors of the defects depth were less than 5%. The experimental results showed that the parameters determined by MOEA could satisfy the optimum parameters in the experiments, and it could provide parameters for project application.