Application Of Wavelet Entropy And Wavelet Networks In Measurement Of Multiphase Flow Parameters

Two-phase flow is frequently encountered in many industry phenomenons. Study of the behavior of flow has important scientific value and the value of a wide range of engineering applications. Two-phase flow measurement has become the hot topic to be explored by a domestic and foreign scientific and technical worker.The development of Tomography Technology and the Soft-sensing Technology provide a new way of thinking and methods for the solution of parameters detection question in multi-phase flow system, and it has been given extensive attention by researchers at home and abroad. Wavelet entropy can reveal information of energy distribution and the changing of characteristic parameters about the data in time- frequency space. In recent years, it has been widely used and studied in engineering. The wavelet network is the result which the wavelet analysis theory and the neural network theory unifies, and it also be a hot issues for research and development worldwide. The research in the four areas is the work foundation of the present paper completion.In light of the parametric measuring techniques and flow mechanism of two-phase flow, based on the study of the measurement method of Electrical Resistance Tomography (ERT) sensors in two-phase flow measurement and using wavelet entropy method, the feature extraction and flow pattern identification of the gas-water two-phase flow in horizontal pipe is completed. Based on the principle of V-cone differential pressure measurement and the comprehensive study of wavelet network theory, the soft-sensing of gas mass flow rate of oil-water and gas-water two-phase flow in horizontal pipe was done. In this paper, the main works accomplished are as follows:1. Based on the research results of wavelet network in the domestic and foreign, the topology and the classified research on the wavelet network has been carried on, a new wavelet network structure is proposed and it is the Hybrid Feedback (HF) Elman Wavelet Neural Network. The training algorithm of the Elman wavelet network for Signal Classification and the Hybrid Feedback (HF) Elman Wavelet Neural Network are derived. 2. Based on the study of the measurement method of ERT sensor in two-phase flow measurement, three organization methods of data series of section measurement information are proposed, and the extraction of five kinds of wavelet entropy of cross-section measurement information of gas-water two-phase flow in horizontal pipe is completed under three organization ways, and the analysis result and the comparison conclusion of using it are given, and the flow pattern identification has been achieved by using the Elman Wavelet Neural Network.3. Through the analysis of the principle of V-cone differential pressure measurement and the comprehensive study of wavelet network theory, the measurement model of mass flow-rate with good generalization ability is put forward. The soft-sensing method of mass flow-rate of oil-water and gas-water based on wavelet network model is achieved, and the validity of the model is verified by experimental results.