一种MEMS电容式压力传感器动态迟滞补偿方法
By employing the LSTM-UKF spatiotemporal data fusion method, Bi-LSTM is used to extract hysteresis features and UKF is combined for real-time correction. This solves the hysteresis and nonlinear coupling problems of MEMS capacitive pressure sensors in dynamic environments and achieves high-precision dynamic compensation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JINTIANHONG ENERGY TECH (BEIJING) CO LTD
- Filing Date
- 2026-04-24
- Publication Date
- 2026-07-17
AI Technical Summary
MEMS capacitive pressure sensors suffer from low measurement accuracy due to hysteresis and nonlinear coupling in dynamic measurement environments, and existing technologies struggle to achieve high-precision compensation under rapid pressure changes.
A spatiotemporal data fusion method based on LSTM-UKF is adopted. One-dimensional sensor time series are mapped to a high-dimensional feature space through phase space reconstruction technology. Hysteresis features are extracted using a bidirectional long short-term memory network (Bi-LSTM), and probabilistic correction of real-time observation data is performed by combining unscented Kalman filtering (UKF). A two-layer coupled system of data-driven prediction and physical model correction is constructed.
It achieves high-precision dynamic compensation across the entire temperature range and measurement range, significantly improving the system's robustness and dynamic response speed, and reducing hysteresis errors and noise effects.
Smart Images
Figure CN122409056A_ABST
Abstract
Citation Information
Patent Citations
A Compensation Method for a Silicon Resonant Pressure Sensor
CN120102002B