Intelligent cleaning control method for sleeper mold
The intelligent cleaning system for sleeper molds, which integrates sensor protection and edge computing, solves the problems of sensor data being susceptible to interference and insufficient system reliability. It achieves efficient and accurate cleaning in dusty environments, ensuring the continuity and stability of production.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA ROAD & BRIDGE
- Filing Date
- 2025-09-02
- Publication Date
- 2026-07-21
AI Technical Summary
Existing sleeper mold cleaning systems are susceptible to sensor data interference in dusty environments, leading to inaccurate measurements, insufficient system reliability, production interruptions, and a lack of effective fault response mechanisms. Furthermore, the multi-sensor system suffers from improper handling of data conflicts, inaccurate registration between historical and real-time data, difficulty in deploying neural network models on edge devices, and inconsistent coordinate systems of actuators due to mechanical installation errors, making it impossible for cleaning control to accurately determine the actual state.
A multi-source sensor integrated sealed sensor chamber is adopted, combined with a slit air knife and a piezoelectric vibrator to prevent sensor contamination. An edge computing unit performs data credibility assessment and lightweight neural network inference to generate cleanup paths. A data arbitration mechanism handles sensor anomalies, historical data is reconstructed and spatially registered, model pruning and quantization techniques are used to deploy neural networks on edge devices, and visual calibration and real-time current monitoring are used for precise control.
It significantly improves the system's operational reliability and production continuity in dusty environments, ensures sensor data quality, achieves real-time and accurate cleaning, enhances the system's robustness and stability, reduces model complexity and resource requirements, and improves cleaning efficiency and accuracy.