Calibration system and calibration method of high-g value accelerometer based on deep learning

Abstract

The invention relates to a calibration system and calibration method for a high-g-value accelerometer based on deep learning. Impact test equipment is utilized for efficiently providing impact environments of different impact levels, and the common accelerometer and a high-precision accelerometer are utilized for repeatedly measuring corresponding impact signals and forming a dataset under different impact environments; then based on a deep learning method, the common accelerometer learns from the high-precision accelerometer; finally, through a training depth neural network, hidden features of measuring signals of the high-precision accelerometer are extracted to calibrate the common accelerometer and improve the measuring performance of the common accelerometer. The intelligent calibration method for the high-g-value accelerometer can be used for quickly calibrating and improving the measuring performance of the common high-g-value accelerometer, and can also be used for intelligentrepairing of part of faulted accelerometers.

Description

technical field [0001] The present invention relates to a theory based on deep learning, in particular to a calibration system and calibration method for a high-g accelerometer based on deep learning. Background technique [0002] Shock excitation is a kind of instantaneous energy transfer, and this instantaneous energy transfer often causes huge damage to the system. Shock phenomena widely exist in the design, manufacture and use of products, such as explosive separation during rocket launches and satellite releases, car collisions, bumps and drops during product transportation and loading and unloading, etc. Therefore, in order to evaluate the reliability and survivability of products in shock environments, and to further improve product performance, shock testing technology is particularly important. [0003] Shock testing techniques are necessarily accompanied by the measurement of the shock response. The shock response is usually represented by an acceleration signal ...

AI Technical Summary

Problems solved by technology

In addition, none of these three methods is suitable for situations where the repeatability of the measurement results is poor

[0006] Excluding environmental factors, the error of the sensor itself is mainly determined by factors such as the sensitive principle of the sensor, the material properties of the sensitive element, the structural design, and the manufacturing process. It is very difficult to improve the performance of ordinary accelerometers with quality and manufacturing level

Images