An improved particle swarm optimizer with behavior-distance models and its application in soft-sensor

This paper proposes an improved PSO (particle swarm optimizer) named as BDPSO (behavior-distance PSO). In BDPSO particle changes fly behaviors guided by the optimum of each particle and the optimum found by the whole particles. In this scheme, individuals can adapt themselves more suitable to search for the destination. Some benchmark functions are tested for comparison the performance between BDPSO and standard PSO. The results indicate that BDPSO is able to locate the global optimum more rapidly and accurately than that of PSO significantly. Furthermore, BDPSO is used to train Neural Network to construct an artificial neural network BDPSONN. Then BDPSONN is applied to construct a soft-sensor of gasoline endpoint and compared with PSONN, the results show that BDPSONN performances better than that of PSONN.