煤田火灾的精准探测和识别是煤火高效防治的关键，自然电位法是探测煤田火灾的一种重要方法，然而现有研究未能全面揭示火区自然电位异常演变规律，干扰信号处理及三维反演解释方法研究也未开展，导致不能精准定位火区的三维位置。本项目通过建立火区自然电场理论模型并开展物理模拟，阐明自然电位异常产生的微观机理及宏观特征；构建火区自然电场与温度场的协同变化表征方法，掌握自然电位异常演变规律；采用卷积滤波技术析离压制非煤火因素导致的随机噪声和系统干扰，提高数据信噪比及可靠度；构建地表自然电场与地下三维空间源电流密度的关联性本构方程，提出三维紧致约束反演算法；在火区现场开展实测，结合钻孔资料对反演结果进行验证和误差分析，逆向修正数据处理技术和三维反演算法，提高反演效率及解译精度，最终实现火区的三维精准靶向定位。项目研究成果可为煤火灾害的精准探测提供理论依据和技术支撑，对煤火的高效防治具有十分重要的理论和现实意义。

The accurate detection and identification of coal fires is the key to their effective prevention and control. Self-potential method is an important method to detect coal fires. However, existing research has not fully revealed the evolution of self-potential anomaly in coal fire areas. The interference signal processing technique and 3D inversion method have also not been developed or proposed, which results in the inability to locate the fire accurately in 3D. In this project, theoretical model of the self-potential field that under the influence of multi-geophysical factors is established. Laboratory experiments are also carried out to clarify the microscopic mechanism and macroscopic characteristics of the self-potential anomaly in coal fire areas. A model for synchronizing the self-potential field and temperature field in coal fire area is constructed, based on which the evolution law of self-potential anomaly is grasped. The convolution filtering technique is proposed to separate and suppress the random noise and system interference caused by non-fire factors to improve data signal-to-noise ratio and reliability. Then, the constitutive equation of correlation between surface self-potential field and underground 3D source current density is constructed. A 3D compact constrained inversion algorithm is proposed. Afterwards, field measurement in coal fire area is implemented and the data is used to verify the efficiency of the inversion results. The data processing technique and the inversion code are corrected and strengthened to improve the inversion efficiency and interpretation accuracy. Finally, the accurate stereoscopic targeting of the fire location is realized. The research results of the project can provide theoretical basis and technical support for the accurate detection of coal fire disasters, and have theoretical and practical significance for the efficient prevention and control of coal fires.