To eliminate the periodic fluctuation of strip thickness caused by roll eccentricity, the particle swarm optimization combined with ant colony optimization (PSO-ACO) algorithm is designed to identify and compensate the roll eccentricity. In order to avoid the PSO algorithm falling into local optimum, pseudo-random-proportional rule is introduced to the determination of the population optimal value in PSO for
improving the diversity of population. The update expression of particle position is improved in combination with the update expression of the pheromone concentration, which makes particles pay more attention to the current search information and accelerates the search speed of particles. The simulation experiment results show that the PSO-ACO algorithm has higher accuracy and satisfactory solution speed in the solution of several typical test-functions and off-line roll-eccentricity identification. Through active roll-eccentricity compensation based on the eccentric identification results of PSO-ACO algorithm, the influence of eccentricity on the strip thickness is reduced significantly and then the accuracy of the strip thickness is improved.