Abstract: In response to the urgent need for real-time and precise prediction of geological risks in TBM tunneling of super-long tunnels, this paper explores a geological prediction technology based on the TBM rock-breaking seismic source HSP method, addressing the emerging requirements of intelligence and real-time performance for geological perception technologies to support intelligent construction. By incorporating the theory of seismic interferometry and analyzing the wave field propagation patterns of rock-breaking seismic sources, an intelligent prediction method of "converting noise into source" was developed. Then, a real-time prediction model that integrates parameter presetting, process optimization and hierarchical decision-making was established, enabling continuous detections during TBM tunneling without the need for shutdowns. Furthermore, an analysis technique based on multiple interactive iterations of detection results was proposed, which significantly improves accuracy in identifying geological targets by enhancing reflection energy spectrum signals from unfavorable geological bodies while suppressing false anomalies. The practical applications during the construction of the Daliangshan No. 1 Tunnel demonstrate that the proposed methodology is effective in identifying unfavorable geological bodies ahead of the working face, such as joint-intensive zones and fracture zones, enabling progressively transparent detection of geological information over medium to long distances. The application results verify the applicability and effectiveness of this technology system. The findings of this study not only provide essential technical support for safety control in TBM construction but also offer an effective solution for preventing and managing geological disasters under similar engineering conditions.