【作者】 罗彤；

【导师】 金宁德；

【作者基本信息】 天津大学 ， 检测技术与自动化装置， 2007， 硕士

【摘要】 本文主要针对两相流流型特征的识别问题,提出了一种分析非线性时间序列混沌动力结构突变的检测方法,即动力学指数分割算法。该方法以混沌突变理论为基础,通过相空间重构,计算出同一时间序列的不同区域上的结构突变点左右两部分的动力结构差异,从而认识动力学系统的内在属性特征,指示出任意时刻系统动力学状态随时间的演化趋势。通过选取典型Henon、Lorenz、Rossler、Chens混沌序列,对算法进行数值模拟试验。采用C-C算法和FNN方法确定相空间重构的嵌入维数和延迟时间;然后对影响动力学指数分割算法结果的各个参数进行考察,选择合适的原始时间序列长度、参考窗口宽度、域值大小等参数,使结构突变点与实际突变点偏差最小;同时,对算法进行加噪检验,结果表明混沌动力结构突变检测方法具有较强的抗噪能力。对垂直上升管中气液两相流和油水两相流电导波动信号采用混沌动力结构突变检测方法进行了处理与分析,结果表明:各种流型的动力结构差异呈递减规律,由此进一步认识了两相流流动系统的复杂动力学特性和运动变化及每一种流型随时间的演化趋势,最终达到辨识两相流流型的目的。所以,混沌动力结构突变检测方法可以作为两相流流型特征识别的辅助工具,并且此方法稳定有效。更多还原

【Abstract】 This paper studies identification of two-phase flow patterns and represents a chaotic dynamic structure catastrophe detection method, also called dynamic exponent segmentation algorithm. Based on chaos and catastrophe theory, this method uses the state space reconstruction to calculate the dynamic structure difference of both the left and the right sides of the catastrophic points in different segments of same time series, consequently distinguishing inherent dynamic characteristics and indicating the changing tendency of the dynamic states of a system at random time.Typical chaotic series are selected to do numerical simulation experimentation for the algorithm,such as Henon,Lorenz,Rossler,Chen’s. Using C-C method and FNN method to determine the optimal delay time and the embedding dimension. Parameters that will influence the result of dynamic exponent segmentation algorithm method is observed to find the proper length of original time series, the width of reference window, the threshold value, and to minimize the difference between the structure catastrophic point and the actual one. Then a noise-adding test is made to prove that the chaotic catastrophe detection has better anti-noise capability.Using chaotic catastrophe detection to analyze the conductance fluctuating signals of gas/liquid two-phase flow and oil/liquid two-phase flow in vertical upwards pipe, and the dynamic structure difference of every flow pattern is descending at the same time. The changing regularity of the two-phase is more clearly acknowledged , as well as the complicated dynamic characteristics and its dynamic system’s movement, and the changing tendency of every flow’s pattern at random time. Ultimately two-phase flow pattern is identified. Therefore chaotic dynamic structure catastrophe detection can be applied as a supplement tool for detecting two-phase flow pattern, and is stable and efficient to knowledge flow pattern’s features of gas/liquid and oil/water two-phase flow.更多还原