基于能量解调和遗传优化神经网络的软测量方法研究

发布时间：2018-03-26 22:29

本文选题：油气水三相流　切入点：能量解调　出处：《燕山大学》2015年硕士论文

【摘要】：多相流动体系在工业生产中是非常热门的一个研究方向,其中流型和各个流型下流量的预测是一项重要的研究内容。由于油气水多相流信息信号具有难以预测的流动特性,所以该方向的研究难度比较大。本文利用能量解调算法对油气水三相流信号进行特征提取,采用遗传优化神经网络构建软测量模型为准确预测流型和流量提供了一种可行的方法。首先,针对油气水三相流特征提取难度比较大、准确性不高的问题,本文重点研究了能量解调算法应用的原理,将优化后的算法应用于油气水三相流的特征提取上。该算法能从油气水三相流信号中提取四个特征参数来表征油气水三相流信号的主要特征。此外实验结果表明该算法在特征提取上具有很好的效果。其次,针对油气水三相流流型和流量预测的问题,本文将遗传算法与神经网络算法相结合构建出具有高质量的软测量模型。利用遗传算法来优化神经网络算法的权值和阈值从而达到提高软测量模型准确性的目的。从性能分析上来看该软测量模型的使用能够极大的提升软测量在流型和流量预测上的准确性。最后,基于油田信号的采集、多种特征提取算法和油气水三相流软测量模型,本论文利用能量解调方法为主要特征提取算法并结合其他特征提取方法,构造出的特征向量训练样本,来训练遗传优化的神经网络软测量模型,实验结果充分证明该软测量模型的性能有较大的提升。将该软测量模型应用于流型的识别,然后用该模型对油气水三相流在各个流型下的油相流量和水相流量进行预测,从实验结果中得出该软测量模型具有很好的预测能力。
[Abstract]:Multiphase flow system is a very popular research field in industrial production. The prediction of flow pattern and flow under each flow pattern is an important research content. Because the information signal of oil-gas-water multiphase flow has unpredictable flow characteristics, In this paper, the energy demodulation algorithm is used to extract the characteristics of oil-gas-water three-phase flow signal. Using genetic optimization neural network to construct soft sensor model provides a feasible method for accurate prediction of flow pattern and flow rate. First of all, it is difficult to extract the characteristics of oil-gas-water three-phase flow and the accuracy is not high. This paper focuses on the principle of the application of energy demodulation algorithm. The optimized algorithm is applied to the feature extraction of oil-gas-water three-phase flow. This algorithm can extract four characteristic parameters from the oil-gas-water three-phase flow signal to characterize the main characteristics of oil-gas-water three-phase flow signal. In addition, the experimental results show that the main characteristics of oil-gas-water three-phase flow signal can be represented by the algorithm. The algorithm has a good effect on feature extraction. Secondly, Aiming at the problem of flow pattern and flow prediction of oil-gas-water three-phase flow, This paper combines genetic algorithm with neural network algorithm to construct a soft sensor model with high quality. Genetic algorithm is used to optimize the weights and thresholds of neural network algorithm to improve the accuracy of soft sensor model. In terms of performance analysis, the use of the soft sensor model can greatly improve the accuracy of flow pattern and flow prediction in soft sensing. Finally, Based on oil field signal acquisition, various feature extraction algorithms and soft sensing model of oil-gas-water three-phase flow, this paper uses energy demodulation method as the main feature extraction algorithm and combines other feature extraction methods to construct feature vector training samples. The experimental results show that the performance of the soft sensor model has been greatly improved. The soft sensor model is applied to flow pattern recognition. Then, the model is used to predict the oil-phase flow and water-phase flow of the oil-gas-water three-phase flow under various flow patterns, and it is concluded from the experimental results that the soft-sensing model has a good predictive ability.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TE31;O359

【参考文献】