A wavelet-homotopy hybrid inversion algorithm for inverse problem of ground penetrating radar (GPR) is proposed.Wavelet is applied to separate inverse problem into different scales.In the maximum scale,initial value of the next scale is found by a homotopy method.In other scales,iteration process is carried out to obtain the global minimum point.The algorithm takes advantages of wavelet analysis and homotopy inversion method.It shows high capability of global search.

This paper presents a new version of the generalized pulse-spectrum technique,which is popularly used for solving the inverse problem of differential equations. This version has the property of immunizing noises. The skill of aving computing time is also given in the process of choosing regularign parameters. In the end, the application on the inverse problem is seismic prospection is discussed, the numerical result of simulation is given.

In this paper, Discrete Newton Method for solving nonlinear operator equations and Tikhonov's regularization method for solving linear ill-posed problems are combined into a stable iterative scheme for solving nonlinear operator equations with hard computed and ill-conditioned derivatives. The results of convergence analyses and error estimates are ob-tained. The application of this method for solving the inverse problem on remote sensing of the thermal conductivity in the nonhomogeneous material is demonstrated. It is shown by numerical simulations that this method is feasible and effective.