射流表面清洁与活化在线监测和反馈控制方法及智能控制系统

By using online monitoring and feedback control methods, multidimensional data is collected in real time and pre-trained neural networks are loaded for adaptive adjustment, which solves the problem of insufficient self-optimization capability in jet surface treatment methods and realizes stable and efficient treatment of jet surface cleaning and activation.

CN122085704BActive Publication Date: 2026-07-17SHANDONG UNIV

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SHANDONG UNIV
Filing Date
2026-04-22
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing jet surface treatment methods lack continuous learning and self-optimization capabilities, and system parameters cannot be adaptively adjusted, resulting in a narrow process window, large fluctuations in yield, and difficulty in achieving stable and consistent high-quality standards.

Method used

An online monitoring and feedback control method for jet surface cleaning and activation is adopted. By collecting multi-dimensional data in real time, a working condition feature encoding vector is constructed, a pre-trained neural network is loaded for online backpropagation training, an adaptive quality threshold is generated, and multi-parameter collaborative regulation is achieved. Combined with global compliance optimization, the model is ensured to operate robustly under dynamic working conditions.

Benefits of technology

It achieves fully automated, highly reliable, and sustainable online assurance of the quality of jet surface cleaning and activation, improves the system's adaptability and stability, and ensures the reliability and consistency of high-end manufacturing processes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122085704B_ABST
    Figure CN122085704B_ABST
Patent Text Reader

Abstract

射流表面清洁与活化在线监测和反馈控制方法及智能控制系统,涉及表面处理技术领域,解决现有射流表面处理方法存在的持续学习与自我优化的能力欠缺、系统参数无法自适应调整的问题。包括数据采集、数据处理、模型更新、自适应阈值决策、反馈控制和全局合规性优化步骤。本发明采用预训练基础模型结合环形缓冲区增量样本集的更新策略,通过冻结基础模型的特征提取层权重,确保已经学习到的射流形态通用特征模式不被破坏;同时解耦全连接输出层并初始化自适应权重矩阵,使模型能够专注于学习新工况下的特异性变化;并且严格约束输出层权重更新范围,确保每次增量训练仅允许参数在小幅区间内动态迁移,避免了因持续学习而导致的灾难性遗忘问题。
Need to check novelty before this filing date? Find Prior Art