重磁联合反演是对重磁数据进行定量解释的重要方法，其特点为多参数、多极值以及非线性。以往将非线性反演问题进行线性化这种做法使反演结果极度依赖初始模型的选取，而基于随机搜索的非线性反演方法虽可以提升反演的全局性，降低对初始模型的依赖度，但是存在正演计算量大，收敛速度慢等问题。本申请拟首先采用控制体积有限元法实现重磁场数值模拟，并基于搜索方向记忆的自适应差分进化算法，实现确定性搜索方向和随机搜索的有机结合，避免不必要的计算量，提升算法的收敛速度。由于重磁深部场源的信息较弱，增加了深部反演的不确定性和解释难度，而地震波场提供的深部密度信息和重力场之间具有一定的联系，因此，将地震速度模型作为约束实现重磁场的联合反演，能有效提高重磁反演的深部分辨率，提升反演结果的可靠性。通过以上研究，在地震数据约束下对青藏高原冈底斯带的重磁数据进行联合反演，旨在验证方法的可行性。

Gravity and magnetic joint inversion which characteristics are multiple parameters, multiple extremes, and nonlinearity is an important method for quantitative interpretation of gravity and magnetic data. Linearization of nonlinear inversion problems in the past has made the inversion results extremely dependent on the selection of the initial model, however, the nonlinear inversion method based on random search can improve the globality of inversion and reduce the dependence on the initial model, but there are problems such as large amount of forward calculation and slow convergence. This project firstly uses the control volume finite element method to realize the numerical simulation of the gravity and magnetic field. Then we propose an adaptive differential evolution algorithm based on search direction memory to realize the combination of deterministic search direction and random search, while this method avoids unnecessary calculations and improves the convergence speed of the algorithm. Due to the weak information of the deep source in the gravity and magnetic field, it’s difficult to inverse the deep information and explain the data. While there is a certain relationship between the density information provided by the seismic field and the gravity and magnetic field. Therefore, using the seismic data information as a constraint to perform the joint inversion of the gravity and magnetic field can effectively enhance the deep resolution improve the reliability of the inversion results. Through the above research, under the constraint of seismic data, joint inversion of the gravity and magnetic data of the Gangdese belt on the Qinghai-Tibet Plateau is carried out to verify the feasibility of the method.