基于粒子群优化极限学习机及电容层析成像的两相流流型及其参数预测

提出了一种基于粒子群优化极限学习机及电容层析成像的两相流流型辨识及其参数预测方法。首先,通过粒子群优化极限学习机的连接权值,并使用粒子群优化极限学习机算法对4种典型的油-气两相流流型进行辨识;其次,使用粒子群优化极限学习机算法对流型的参数进行预测;最后进行了仿真实验,结果表明,与极限学习机算法相比,粒子群优化极限学习机算法所需隐层节点数更少,流型辨识率更高,其正确辨识率达100%,对4种流型参数预测的最大相对误差为5.24%。

Two-phase Flow Regime and its Parameter Prediction Based on Particle Swarm Optimization Extreme Learning Machine and Electrical Capacitance Tomography

Prediction method for two-phase flow regime and its parameter based on particle swarm optimization extreme learning machine (PSO-ELM) and electrical capacitance tomography is presented. Firstly, the weights of extreme learning machine are optimized using particle swarm optimization algorithm, and then particle swarm optimization extreme learning machine algorithm is adopted to identify four typical oil-gas flow regimes. Secondly, the parameters of the four flow regimes are predicted by particle swarm optimization extreme learning machine algorithm. Finally, simulation experiments are carried out and the results show that particle swarm optimization extreme learning machine algorithm needs less hidden layer nodes and has higher accuracy for flow regime identification compared with extreme learning machine algorithm. The correct identification rate is 100%, and the maximum relative error for the four flow regimes is 5.24%.