Development of a hybrid particle swarm optimization algorithm for multi-pass roller grinding process optimization

In the field of metal rolling, the quality of steel roller’s surface is significant for the final rolling products, e.g., metal sheets or foils. The surface roughness of steel rollers must fall into a stringent range to guarantee the proper rolling force between the sheet and the roller. To achieve the surface roughness requirement, multiple grinding passes have to be implemented. The current process parameter design for multi-pass roller grinding mainly relies on the knowledge of the experienced engineers. This always requires time tedious “trial and error” and is insufficient to work out cases: (1) multi-pass with complex interaction for one pass with its neighboring passes; (2) large number of process parameters setup; (3) multiple process objectives and constrains. In this paper, a process planning method for multi-objective optimization is proposed with a hybrid particle swarm optimization while incorporating the response surface model of the surface roughness evolution. The hybrid particle swarm optimization regards the entire grinding process parameters (from rough grinding, semi-finish grinding, finish grinding to spark-out grinding) as a whole, and realizes the parameter optimization by considering multiple objectives and constrains. The establishment of the response surface model of surface roughness evolution is capable to incorporate the inter-correlation of neighboring passes into the multi-pass parameter optimization. Finally, the experimental verification was implemented to verify the effectiveness of the proposed method. The error between predicted roughness and experimental roughness is less than 16.53%, and the grinding efficiency is improved by 17.00% compared with the empirical optimal process parameters.