复杂重磁异常多阶段决策的最优化反演

Optimum inversion of sophisticated gravitational and magnetic anomalies by multi-stage decision.

用常规的阻尼最小二乘法和广义逆矩阵法反演由多个地质体产生的复杂重磁异常 ,往往不收敛。采用非线性规划的多阶段决策方法对目标函数作小波多尺度分解 ,由大尺度到小尺度逐次反演 ,消除了目标函数的冗余度 ,并克服了目标函数的复杂性。实践表明 ,多阶段决策方法能解决由多个地质体产生的复杂的多峰值异常的最优化反演问题 ,并且通过多尺度分解目标函数来搜索极小的过程是合理和收敛的。

The unconvergence often occurred during the inversion of sophisticated gravitational and magnetic anomalies produced by several geologic bodies when conventional damped least square inversion or generalized inverse matrix inversion has been used. Using the method of multi stage decision based on non linear programming to do multi scale wavelet decomposition of objective function and doing successively inversion from large scale to small scale, the redundance of objective function has been overcome. The practice proved that the method can solve the optimum inversion of sophisticated anomalies with multi peak generated by multi geologic bodies, and its process of searching minimum through decomposing the objective function by multi scale is reasonable and convergence. The method has been used to process and interpret the distribution of igneous rock in a certain oilfield.