Optimization of Sour Water Stripping Unit Using Artificial Neural Network–Particle Swarm Optimization Algorithm

Sour water stripping can treat the sour produced by crude oil processing, which has effect of environmental protection, energy saving and emission reduction. This paper aims to reduce consumption unit strengthening process parameter optimization. Firstly, basic model is established utilizing Aspen Plus, optimal determined comparative analysis back propagation neural network (BPNN), radial basis function (RBFNN) generalized regression (GRNN) models. Then, sensitivity Sobol used select operating variables that have a significant influence on system. Finally, particle swarm optimization (PSO) algorithm optimize conditions unit. The results show RBFNN more accurate than other Its structure 5-66-1, expected value an approximately linear relationship with output value. Through analysis, it found each impact process, needs be optimized PSO algorithm. After 210 iterations algorithm, system obtained. In addition, cold/hot feed ratio, sideline production position, tower bottom pressure, hot temperature, cold temperature are 0.117, 18, 436 kPa, 146 °C, 35 respectively; 5.918 MW. Compared 7.128 MW before optimization, greatly reduced 16.97%, shows energy-saving very significant.