一种胶原蛋白提取设备状态监测方法及装置

By employing a multi-sensor system and a slice recombination strategy guided by dynamic time warping, combined with a frequency domain fusion mechanism based on learnable wavelet bases and dynamic feature selection, the problem of insufficient correlation and synergistic characteristics of sensor data in existing collagen extraction equipment monitoring is solved, enabling accurate monitoring and robust prediction of equipment status.

CN121655916BActive Publication Date: 2026-07-17SHANDONG HENGXIN BIOTECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SHANDONG HENGXIN BIOTECH CO LTD
Filing Date
2025-12-10
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing monitoring methods for collagen extraction equipment rely on single or a small number of sensors, which cannot fully reflect changes in equipment status. They also lack the temporal correlation and cross-sensor synergistic characteristics of multi-sensor data, resulting in inaccurate monitoring results and insufficient model robustness.

Method used

A multi-sensor system is used to collect four types of time-series data in real time. Through a slice recombination strategy guided by dynamic time warping and a frequency domain fusion mechanism with learnable wavelet basis, an adaptive frequency domain feature fusion module, a cross-modal feature cross-enhancement module, a dynamic feature selection gating mechanism, and a multi-scale convolutional feature extraction module are constructed. Combined with a state-aware gated recurrent unit and a state probability prediction module, accurate monitoring of equipment status is achieved.

Benefits of technology

It improves the accuracy and robustness of equipment condition monitoring, effectively captures complex feature patterns of equipment, dynamically adapts to state changes under different operating conditions, and enhances the efficiency of frequency domain feature utilization and the interpretability of feature extraction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121655916B_ABST
    Figure CN121655916B_ABST
Patent Text Reader

Abstract

本发明涉及一种胶原蛋白提取设备状态监测方法及装置，属于人工智能技术领域。其包括以下步骤：获取胶原蛋白提取设备监测数据；获取胶原蛋白提取设备静态参数；通过计算自适应切片长度和动态切片数，将多传感器时序监测数据重组为三维张量；构建设备状态监测模型；采用交叉熵损失作为主损失项计算模型预测的设备状态概率分布与真实标注标签之间的差异；采用小批量梯度下降策略对模型进行训练，得到训练好的模型；对新采集的时序监测数据进行处理生成重组三维张量后与设备静态参数输入到训练好的模型中，输出设备状态概率分布，判定设备状态，进行预警。本发明能够提升模型的分类精度和鲁棒性。
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Sand separator monitoring control method based on multi-source data processing

    CN120316719A

  • Support structure stress state monitoring method based on artificial intelligence

    CN120524256A