Velocity sharing historical best particle swarm optimization algorithm for soft sensor modelling in ammonia synthesis process

In this study, a velocity sharing historical best particle swarm optimization algorithm (VSHBPSO) and its variants are proposed to improve the performance of the original particle swarm optimization (PSO). The shared information in the improved algorithms includes the historical best position of each particle searched in the previous experiments, the updated velocity and the present global best position. An orthogonal design trial is conducted to discuss the parameters of the proposed algorithms by using 10 non-linear functions with different dimensions. Furthermore, the performance of the new algorithms is evaluated. Experimental results show that the novel algorithms can derive better solutions than the PSO algorithm and indicate their effectiveness in optimizing non-linear functions. Finally, the proposed algorithm is applied in soft sensing the outlet ammonia content in the ammonia synthesis process. The VSHBPSO-based soft sensor is found to be effective in the real-time assessment of ammonia content.