A Novel Parallel Processing Model for Noise Reduction and Temperature Compensation of MEMS Gyroscope

Area of Science:

• * Instrumentation and Measurement Science
• * Signal Processing
• * Control Systems Engineering

Background:

• * Micro-Electro-Mechanical Systems (MEMS) gyroscopes are crucial for navigation and motion sensing.
• * Output signals are susceptible to noise and temperature drift, limiting measurement accuracy.
• * Existing compensation methods often struggle with complex signal interferences.

Purpose of the Study:

• * To develop a parallel processing model for accurate MEMS gyroscope signal compensation.
• * To effectively eliminate noise and temperature drift from gyroscope output signals.
• * To improve the overall measurement accuracy and stability of MEMS gyroscopes.

Main Methods:

• * Variational Mode Decomposition (VMD) optimized by Multi-objective Particle Swarm Optimization (MOPSO) for signal decomposition.
• * Classification of intrinsic mode functions (IMFs) into noise, mixed, and drift segments using Sample Entropy (SE).
• * Parallel processing involving direct noise segment discarding, Time-Frequency Peak Filtering (TFPF) for denoising, and Beetle Antennae Search algorithm (BAS)-optimized Elman Neural Network (Elman NN) for drift compensation.

Main Results:

• * Significant reduction in angle random walk from 0.531076 to 5.22502 × 10^-3°/h/√Hz.
• * Substantial decrease in bias stability from 32.7364°/h to 0.140403°/h.
• * Demonstrated superiority of the proposed parallel processing model in enhancing gyroscope performance.

Conclusions:

• * The MOVMD-TFPF and BAS-Elman NN parallel model effectively compensates for MEMS gyroscope noise and temperature drift.
• * The proposed method significantly improves gyroscope accuracy and stability.
• * This approach offers a robust solution for high-precision inertial measurement applications.