ECT Image Reconstruction Algorithm Study Based On Bp Neural Network

ECT is a kind of important nondestructive testing technology, because of its non-contact, flexibility, and the advantage of high precision. Especially in the field of two phase flow detection, the technology is able to reflect the characteristics of the flow pattern by 2D and 3D image more effectively.The nonlinearity and morbid state existed in ECT image reconstruction reduces the image quality and real-time quality. BP neural network is applied as a ECT image reconstruction algorithm, implementing the direct nonlinear mapping between measured capacitances and pixel values of reconstruction image. three tasks are listed as follows:1.Solving positive problems.Simulating the sensitive field in COMSOL and solving the model.2.Reseach on Image Reconstruction Algorithms for ECT.Common principle of the algorithms are further studied and implemented by programming. Improved Image Reconstruction Algorithms are put forward. Compared with traditional algorithms, BP neural network avoids the solution of the sensitivity matrix, improving qualities including rate. accuraty and distinguishability.3.Developping the upper machine.Developping a kind of software for Image Reconstruction based on MATLAB. The software consists of following modules: UART module, DSP module, algorithms implement module and image process module. Data is collected by UART module and demodulated. Image enhancement module improve image quality through Suppressing noise.