Octane Number Prediction of Blend Gasoline Based on Improved Particle Swarm Optimization

Abstract

The octane number of hydrogenated gasoline is difficult to be obtained in real time in the modeling of finished gasoline blending formula. Considering the problems of XGBOOST algorithm, gradient lifting tree algorithm and random forest regression algorithm network, a dynamic harmonious search hybrid particle swarm optimization (DSHPHO) algorithm was proposed to predict the octane number of finished gasoline. In this algorithm, the improved HS algorithm is embedded into the PSO algorithm, and all the particles are considered as harmonious memory (HM). Search by harmony search (HS) algorithm of randomness and evolution mechanism to improve the diversity of particle swarm, makes more ergodic particle swarm at the beginning of the search, reduce sensitivity to the initial value of the algorithm and keep randomly generated in the whole evolution process of the possibility of new particles, fundamentally solves the particle swarm optimization algorithm in dimension increase diversity is less defects. The algorithm has faster convergence speed and better global search ability. Finally, based on this method and industrial historical data, the octane number prediction model of hydrogenated gasoline components is established. The simulation results show that the dynamic harmonious search hybrid particle swarm optimization algorithm has better prediction performance than the traditional particle swarm optimization algorithm, and can be used to predict the octane number.