铀矿勘查中壤氡测量技术及异常信息提取方法研究

The exploration of uranium deposits keeps the champion status in nuclear energy industry and the resources of uranium is the strategic priority of China’s resource exploration. It is proved that the method of measuring radon in soil is an effective technical measure for uranium exploration .This project aims at the technical demand for exploring deep deposit in uranium exploration. The experimental data and environmental parameters of the corresponding parameters of soil radon were obtained by establishing the experimental system of real-time accumulation of multi-parameter measuring system of arrays about soil radon in the proven uranium mining area. Firstly, the soil radon data of each array are normalized by the neural network method. Then, With the character of the signal local features both in frequency and time domain shown in the Gaussian mixing up GMM statistical decision analysis and multi-resolution analysis method. Finally, According to different frequency characteristics, Separate the array of soil radon measurements. Extract the distribution characteristics of soil radon data, and the higher order moments formed the eigenvectors in the sub – band. By adopting the principal component analysis to find the main features of the large dispersion among other data; Obtain characteristic information about the "cumulative radon" from the earth’s surface and the " instantaneous radon " from the deep underground. And then extract the anomalous information of the earth’s surface soil radon caused by deep Uranium deposit and explore the correlation between soil radon anomaly and deep uranium deposit, to provide effective measurement technology of soil radon and data processing method for the new round of uranium exploration.

铀矿勘探是核能产业的龙头，铀矿资源是我国资源勘查战略重点，实践证明壤中氡气测量方法是铀矿勘查中的一种有效技术手段。本项目针对铀矿勘查中攻深找盲的技术需求，通过在已探明铀矿区地表建立阵列式壤氡α能谱累积多参数测量的实验系统，获得相应的壤氡α能谱累积多参数和同步的壤氡α杯累积实验数据及环境参数。先通过神经网络方法归一化各阵列测点的壤氡数据；然后通过高斯混和GMM统计决策分析及多分辨率分析方法，利用其在时域频域都具有表征信号局部特征的能力，将阵列式壤氡测量数据按不同频率特征分离，在子频带提取壤氡数据分布特征、提取高阶矩构成特征向量；通过采用主分量分析方法寻找与其它数据存在较大分散度的主要特征；从而获得来自地表的“累积氡”和来自地下深部的“瞬时氡”特征信息；进而提取深部铀矿引起的地表壤氡异常信息，探索壤氡异常与深部铀矿床关系的相关规律，为新一轮铀矿勘查攻深找盲提供有效的壤氡测量技术和数据处理方法。

铀矿勘探是核能产业的龙头，铀矿资源是我国资源勘查战略重点，实践证明壤中氡气1 测量方法是铀矿勘查中的一种有效技术手段。本项目针对铀矿勘查中攻深找盲的技术需求，通过在已探明铀矿区地表建立阵列式壤氡实时累积多参数测量的实验系统，获得相应的壤氡累积多参数和壤氡α杯累积实验数据及环境参数。先通过神经网络方法归一化各阵列测点的壤氡数据；然后通过高斯混和GMM统计决策分析及多分辨率分析方法，利用其在时域频域都具有表征信号局部特征的能力，将阵列式壤氡测量数据按不同频率特征分离，在子频带提取壤氡数据分布特征、提取高阶矩构成特征向；通过采用主分量分析方法寻找与其它数据存在较大分散度的主要特征；从而获得来自地表的“累积氡”和来自地下深部的“瞬时氡”特征信息；进而提取深部铀矿引起的地表壤氡异常信息，探索壤氡异常与深部铀矿床关系的相关规律，为新一轮铀矿勘查攻深找盲提供有效的壤氡测量技术和数据处理方法。