基于微粒群优化算法的人工神经网络模型预测汽液相平衡常数

Applying Particle Swarm Optimization Algorithm-based Artificial Neural Network to Predict Vapor-Liquid Equilibrium Constants

利用微粒群优化算法对人工神经网络进行训练,克服了BP网络收敛速度慢,易陷于局部最优解的缺点。用此网络模型来预测汽液相平衡常数,并利用严格模型计算的多组相平衡数据作为训练样本和检验样本来检验利用微粒群算法训练的人工神经网络。结果表明,此种方法收敛速度快,精确度高,好于传统的BP算法。

Particle swarm optimization (PSO) algorithm was used to train the artificial neural (ANN) network. This method could overcome the drawbacks of slow convergence and easy to locate in local-optimum which always happened when using BP network. The PSO-based ANN was used to predict the vapor-liquid equilibrium constants and the results showed that this method was much better than the traditional back propagation algorithm for both the rate of convergence and the precision .