A predictable artificial physics optimisation algorithm (Q2224045)

Summary: The paper develops a predictable artificial physics optimisation (PAPO) with the aim to improve the global performance of APO. PAPO emphasises the prediction of individual movement in a continuous movement process according to the historical information by PD controller. The APO system is regarded as a two-order dynamic system. The architecture of APO system is constructed based on z-transformation. The PD controller is introduced in the feedback channels of the architecture of APO system to control the individuals dynamically, which can prompt the individuals responding to the change of its own history movement correctly and rapidly. In PAPO, individuals always make predictions for the future position according to their own inertia motion position in the flight process, and then adjust their velocity according to the distance between the prediction position and the swarm weighted position. Due to emphasise the continuity and inertia of the particle's own movement, the new model describes individual motion behaviour more accurately than APO system. Simulation results show PAPO algorithm can improve the population diversity and global search capability of APO algorithm when solving high dimensional global optimisation problems.