【摘要】 本文提出了基于BP(Back Propagation)神经网络和列队竞争算法的方法优化低温甲醇洗工艺操作参数。首先,以流程模拟软件对该工艺进行模拟,获得初始样本数据;然后,应用BP神经网络对所获数据进行训练,实现神经网络输出与实际模拟结果的一致;最后,综合列队竞争算法和BP神经网络对低温甲醇洗工艺过程进行参数优化。实例计算结果表明,训练好的BP神经网络输出与实际模拟结果间的误差小于2%,将其应用于优化计算可大幅缩短计算时间,提高计算效率;优化计算结果能够在满足分离要求的条件下,降低公用工程消耗量13%,降低气提用氮气量8.1%,节能效果十分明显。更多还原

【Abstract】 A novel method based on line-up competition algorithm and BP neural network is proposed in this paper for solving the optimization of parameters in rectisol wash process.First,the rectisol process is simulated by the process simulator to achieve enough data points.Second,the BP neural network is applied to the training of those data.The outputs of the trained BP neural network is close to the results of the process simulation.At last,the line-up competition combined with the BP neural network is used in the optimization of the rectisol process.The casestudy results indicate that the error between the outcome of the trained BP neural network and that of the Aspen Plus is less than 2%.Computing with the BP neural network can decrease the computation time dramatically and improve the efficiency significantly.The utilities consumption can be decreased by 13%and the N₂ consumption in stripping can be reduced by 8.1%after optimization,which has an obvious benefit on energy consumption.更多还原