Geothermal fluid source discrimination method based on geochemical index and machine learning fusion
By integrating geochemical indicators with machine learning to identify the source of geothermal fluids, an end-member feature library is constructed. By utilizing attention mechanisms and physical constraints, high-precision quantitative identification and dynamic prediction of geothermal fluid sources are achieved. This solves the problems of insufficient multi-factor coupling analysis and insufficient dynamic prediction in traditional methods, and supports the refined management of geothermal resource exploration.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-19
- Publication Date
- 2026-03-27
AI Technical Summary
Existing technologies for identifying the source of geothermal fluids suffer from insufficient ability to analyze multiple factors coupled together, ambiguity in the interpretation of isotope indices, and a lack of dynamic evolution prediction capabilities, making it difficult to meet the complex needs of geothermal resource exploration and development.
A geothermal fluid source identification method based on the fusion of geochemical indicators and machine learning is adopted. Through the acquisition and feature construction of multi-source geochemical data, an endmember feature library is constructed using unsupervised clustering. A machine learning model with an attention mechanism is trained, and quantitative inversion of mixed proportions is carried out in combination with physical constraints. Finally, a time series prediction model is constructed to achieve dynamic evolution trend prediction.
It has achieved high-precision quantitative identification and dynamic prediction of geothermal fluid sources, solved the problem of multi-factor nonlinear coupling, eliminated ambiguity in isotope interpretation, and provided a basis for refined management decisions throughout the entire life cycle of geothermal fields.
Smart Images

Figure CN121743985A_ABST
Abstract
Citation Information
Cited By
Formation pressure while drilling advanced prediction method and system based on parallel TCN-LSTM-attention and adaptive update
CN122365402A