Research Of “Transparent” Based On 3D Gravity And Magnetic Inversion Modeling

Three-dimensional geological modeling is an important way to achieve breakthroughs in deep mineral exploration.Through three-dimensional modeling of ore-controlling geological bodies,the spatial relationship between ore-controlling elements can be intuitively portrayed and displayed,so as to realize the "transparency" of the mining area,which helps Understand and understand the metallogenic system and conduct deep prospecting and prediction.In this study,the latest measurement of Zhuxi ore field and its periphery in northeastern Jiangxi mining area was used to obtain 1: 50,000 gravity data and 1: 50,000 aeromagnetic data.First,the potential field separation was carried out to obtain anomaly data for gravity and magnetic inversion.Then,the relationship between physical properties and lithology was analyzed.Under the constraints of surface geology and borehole data,human-machine interactive gravity and magnetic inversion of 17 profiles was completed.Finally,through these 17 profiles,a threedimensional geological model of the Zhuxi deposit and its adjacent area was established.Based on the establishment of the three-dimensional model,the BP neural network algorithm in machine learning is used to train the known mining units as samples,and the mineralization prediction is carried out in Zhuxi area.The thesis has obtained the following main understandings and achievements:(1)Obtained a three-dimensional geological-geophysical model around Zhuxi in the northeastern Jiangxi mining area,obtained the three-dimensional spatial morphology of the rock mass and stratum at a depth of 5km underground,and initially achieved "transparency"(2)Use BP neural network to make prospecting predictions for the ore-concentrating area,and obtain the distribution maps of the favorable degree of metallogenic formation in different positions of the ore-concentrating area,including 15 most favorable metallogenic units,the distribution position is divided by the predecessors The prospects for mineralization are generally consistent and there are new discoveries.(3)Human-computer interaction and inversion are carried out under the knowledge of physical property parameters and other information,so that the model obtained by modeling is more in line with the real situation,and the inversion accuracy is improved;the use of BP network for mineralization prediction can integrate multi-source data,Make full use of prior information and key ore-controlling factors to improve the accuracy of mineralization prediction.Through the "transparency" study of the Zhuxi orefield in northeastern Jiangxi,and digging deep into the gravity and magnetic information,we have a better understanding of the deep geological structure of the study area,such as the fault structure,the stratum distribution of the study area,the distribution law of the rock body,etc..It provides a basis for the continuous exploration of the mining areas in Northeast Jiangxi,and provides ideas and techniques for the search for deep metal resources in similar mining areas.With the massive growth of data and the development of data feature extraction in machine learning,the deep integration of artificial intelligence,machine learning and deep resource search and mineralization prediction will greatly improve the prospecting effect and efficiency.