In order to meet the processing needs of various types of products,it is particularly important to improve the highly automatic control ability of forging press.A single intelligent fault diagnosis algorithm cannot achieve high fault diagnosis rate and complex diagnosis measures need to be set.The fault diagnosis was achieved by combining fault tree analysis （FTA） and chaotic particle swarm optimization (CPSO) algorithm.The measurement points of actual sensors were analyzed and compared with the results of GA and PSO algorithms.The results show that the fault diagnosis accuracy of FTA-CPSO algorithm is 6.25% higher than that of GA and 4.20% higher than that of PSO;the FTA-CPSO algorithm can achieve smaller error and better diagnostic performance than GA and PSO;after more iterations,the adaptive value of all algorithms in this experiment decreases;FTA-CPSO can complete the iterative calculation in the shortest time,which can effectively reduce the number of iterations,and the search time is also significantly shortened.