Non-Liner Error Compensation of the New Vibration Accelermeter Based on Neural Network

Abstract

A new type of piezoelectric vibration acceleration sensor was studied, which sensitive element consists of cymbal transducer and a buffer chip. The nonlinear characteristic of voltage sensitivity of this vibration acceleration sensor was investigated. Using the general principles of nonlinear calibration method of BP neural network, established the mathematic model of nonlinear calibration of output voltage sensitivity for this piezoelectric vibration acceleration sensor, and then accomplished the computer simulation programming using MATLAB algorithm. The computer simulation results showed that the nonlinear error of output voltage sensitivity for this piezoelectric vibration acceleration sensor can be less than 0.1‰ after six-steps training using artificial neural network. The effect of the compensation is better than that of commonly piecewise linear interpolation method, and with the characteristic of simple structure and high accuracy.

Keywords:

• Neural network
• Cymbal transducer
• Vibration accelerometer
• Non-linear compensation
• BP algorithm

Acknowledgment

The authors would like to thank the financial support by the Funding Project for Academic Human Resources Development in Institutions of Higher Learning Under the Jurisdiction of Beijing Municipality (PHR200907124).