A PSO-LP Cooperative Algorithm for Mixed Integer Nonlinear Programming

Mixed integer nonlinear programming (MINLP) problems widely exist in various aspects of an oil refinery, from crude oil supply chain, crude oil scheduling, production process decision-making, to optimal design of heat exchanger network or hydrogen network. Due to the NP-hard nature, many MINLP problems are difficult to solve efficiently and reliably. In this paper, a cooperative algorithm based on an improved particle swarm optimization algorithm (PSO) and linear programming algorithm (LP) is proposed to solve the complex MINLP problems. The PSO-LP cooperative algorithm includes an outer module running the improved PSO algorithm, and an inner module running LP algorithm, for example, the simplex algorithm. The outer module transforms the MINLP problem into a LP problem and passes it on to the inner module. The inner module feeds back the LP results to the outer module to update the MINLP problem. The above procedure iterates continuously to get the optimal solution of the MINLP problem. The proposed PSO-LP cooperative algorithm is verified using various MINLP problems. Results show that the proposed algorithm is computationally more accurate and more reliable than the existing ones.