Multi-objective optimization of propane recovery process based on improved BP neural network

Based on the actual operating data of a processing plant, the direct heat exchange process(DHX) of propane recovery under different operating parameters is simulated by using the process simulation software HYSYS. This paper analyzes the influence of low temperature separation temperature, DHX top temperature and reflux tank temperature on propane yield and system energy consumption. An improved BP neural network is used to establish a multi-objective optimization model of the process, and NSGA-Ⅱ (non-dominated sorting genetic algorithm) is used to solve the multi-objective solution. The results show that the improved back propagation(BP) neural network has high prediction accuracy for propane yield and system energy consumption, and the relative errors are all below 2%. The Pareto solution set obtained by the NSGA-Ⅱcan provide a guiding role for process design and actual production.