Method Of Flow Pattern Identification Based On Electrical Capacitance Tomograph (ECT) System

Electrical capacitance tomography (ECT) is a process tomography(PT) technique based on capacitance sensor. ECT has the advantage of being non-radiate, non-intrusive, fast in response, simple in structure and low in cost, so it is a new way to measure two-phase flow parameters. Flow pattern identification is a important question in the measurement of two-phase parameters which can be resolved with ECT technology, which is a good potential, in process of groping approach. This paper makes complete and embedded research on methods of flow pattern identification, and brings forward neural network with feature extracted to distinguish flow pattern, at the same time validates and estimates this method with experimented data from experimentation. The main contribution and results are as follows:1. It expatiates the basic principle, system construct of ECT , and researches actuality of flow pattern identification.2. The ECT experiment board is debugged; the software of singlechip is programmed; some arisen questions in debugging are discussed.3. Based on finite element theory, capacitance sensor model is built, and variable medium distributings are set to simulate flow pattern, and capacitance values are calculated.4. It analyses simulation data in search of parameters which can delegate pattern information, extracts ten parameters in order to distinguish pattern effectively, simply, fast and exactly.5. Some methods of flow pattern identification is synthesised, and it brings forward compete neural network and BP neural network based on distinguishable feature extracted straight from the capacitance measurements to distinguish flow pattern. Two methods indicate that distinguishable feature can simple network construction, decrease train difficulty and increase veracity.6. Video of flow pattern is realized with the linearity back projection(LBP) arithmetic .7. 8-pole capacitance sensor is made with 50mm pipeline caliber which can fill material. Data is acquired from experimentation and the feasibility and reliability of the method is validated which identifies flow pattern. The result shows that the neural network based on feature exctracted is effective.