Quasi-Newton Method in Electrical Resistance Tomography for Measurement of Two-phase Flow

摘要

Electrical Resistance Tomography (ERT) is one of new methods for parameters measurement of two-phase flow. The image reconstruction algorithm is one of key parts of ERT system. Based on the optimization principle of BFGS algorithm (one kind of the Quasi-Newton methods), a new image reconstruction algorithm of ERT for two-phase flow is proposed. To satisfy the requirement of real-time measurement of two-phase flow, the new algorithm (the modified BFGS algorithm), which results in computation decreasing and image reconstruction speed rising, selects an inexact search direction and the Hessian matrices of the objective function are replaced with identity matrix. Real-tune experiments show that the reconstruction speed of the modified BFGS algorithm is slightly slower than that of the commonly used sensitivity coefficient algorithm. However, the quality of images reconstructed by the modified BFGS algorithm is far higher than that of images obtained by the sensitivity coefficient algorithm.