Geological Section Classification With Knowledge-Based Feature Engineering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for classifying geological sections fail to fully leverage geological knowledge, resulting in low accuracy due to complex nonlinear relationships and lack of professional knowledge integration.
Innovation Solution
A method involving on-site exploration data and remote sensing data preprocessing, feature extraction and construction, selection, and normalization, followed by constructing a geological section classification model using LightGBM as the benchmark architecture, incorporating geological features and new features to enhance model performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning models use only raw data and basic features, then the model complexity is low, but the classification accuracy deteriorates due to inability to capture complex nonlinear relationships and lack of geological knowledge integration
Solution Approach 1:
The patent applies preliminary action by pre-processing exploration data and remote sensing data before feeding them to the model. This includes data cleaning, normalization, and feature extraction to create high-quality input features that capture geological patterns, thereby improving model accuracy without requiring overly complex model architectures.
Solution Approach 2:
The patent introduces an intermediary layer of geological knowledge integration between raw data and the classification model. This involves incorporating geological domain knowledge into feature engineering and model design, acting as a mediator that bridges raw data and the model to achieve better classification accuracy.
2Measurement precision
If geological knowledge is fully integrated into the classification model, then the classification accuracy improves, but the ease of operation deteriorates due to increased model complexity and knowledge requirements
Solution Approach 1:
The patent applies segmentation by dividing the complex classification task into multiple stages: data pre-processing, feature extraction, feature selection, and classification. This modular approach integrates geological knowledge at each stage while maintaining operational simplicity through clear separation of concerns and standardized interfaces between stages.
3Measurement precision
If complex nonlinear relationships in geological features are considered, then the classification accuracy improves, but the device complexity increases due to more sophisticated model architecture requirements
Solution Approach 1:
The patent applies parameter changes by transforming raw geological features into optimized feature representations through pre-processing and feature engineering. This includes applying mathematical transformations, normalization, and selection of key parameters that capture nonlinear relationships while maintaining compatibility with standard classification algorithms.
Data Source
AI summary
The disclosure relates to the technical field of geological exploration, and disclosed in the disclosure are a method for classifying a geological section, a storage medium and a device. The method includes: acquiring on-site exploration data and remote sensing data of a geological section, and a category label of the geological section, preprocessing the data, and extracting features; constructing new features on the basis of extracted features, and forming a feature set; selecting and normalizing features in the feature set, and dividing same into a training set and a validation set; and constructing a geological section classification model, and inputting data to be tested into the model after validating to obtain geological section classification results. According to the method of the disclosure, knowledge related to geology is fully utilized to enhance feature expression ability, improve generalization ability and interpretability of the model, and capture complex nonlinear relationships.