CSAMT Electrical Feature Enhancement and Classification Method Based on Machine Learning in Deep Mined-Out Area

Abstract

The invention discloses a method for enhancing and classifying electrical characteristics of CSAMT in deep mined-out areas based on machine learning. Convolution kernel of goaf features; II. Use the convolution kernel obtained in step I to perform convolution calculation to extract shallow anomaly information and deep gob electrical anomaly information; III. Shallow convolution obtained in step II Calculate the resistivity feature and the deep goaf convolution resistivity feature to obtain the energy error. If the energy error does not meet the accuracy requirements, return to step I to modify the initial convolution kernel parameters; if the energy error meets the error requirements, remove the shallow part. Abnormal interference, enhance the electrical characteristics of deep mined-out areas; IV. Carry out fuzzy clustering analysis based on kernel function on the electrical characteristics of CSAMT obtained in step III; V. According to the clustering results, conduct Identify and classify, and predict the basic situation of gobs.

Description

technical field [0001] The invention relates to a machine learning-based CSAMT electrical feature enhancement and classification method for deep mined-out areas. Background technique [0002] Mined-out areas are likely to induce accidents such as mine earthquakes, water inrush, and toxic and harmful gas leakage, which seriously threaten the safety of mine personnel and property. At present, the methods for detecting gobs mainly include transient electromagnetic method, high-density electrical method, detection radar method, Controlled Source Audio-frequency Magnetotellurics (CSAMT) and so on. [0003] The above methods have achieved good prediction results. Among them, CSAMT has the advantages of large exploration depth range, strong anti-interference ability, small high-resistance shielding effect, high resolution, and high work efficiency. It is widely used in geothermal resource detection, bridge and tunnel engineering, etc. , coal mine goaf prediction, hydrogeology, cav...

AI Technical Summary

Problems solved by technology

[0004] In recent years, although CSAMT has achieved good results when applied to gob detection, however, due to the strong low-resistivity anomalies that may be caused by surface rivers, ponds, and near-surface groundwater collection areas, these low-shallow low-resistivity Abnormalities will affect the inversion accuracy of CSAMT and reduce the electrical characteristics of deep goafs, so it is impossible to accurately identify and classify goafs, roadways, etc., so that the location, distribution range, scale, etc. base case forecast

Images