Abstract
Current optimisation methods, especially stochastic methods, are short of intermediate data analysis and control. Moreover the intermediate data has rarely been interpreted into the knowledge useful for either optimisation itself or industry productions. In this paper, we proposed a new optimisation method featuring on the capability of intermediate data analysis and control without setting any computational burden on optimisation. The method makes use of the long Markov process whose concept is borrowed from that in the simulated annealing but meanwhile bypasses its inherent sequential nature by using the proposed conceptual pools to populate the generated solutions. Four simple optimisation problems were selected to investigate the validity of the new algorithm. The algorithm was also studied on a complex engineering problem by comparing with the simulated annealing. Results showed that the new algorithm was as robust as the simulated annealing algorithm but can achieve quicker convergence.