Aiming at temperature drift problem of torsion silicon micro-accelerometer

a back propagation（ BP）neural network temperature compensation method optimized by adaptive weight particle swarm optimization（ PSO）is proposed based on comparative analysis of compensation effects of different algorithms. At the same time

by referring to the idea of variation in traditional genetic algorithm

the introduction of mutation in PSO algorithm overcomes the defect that BP neural network is easy to fall into local extremum

and the precision of temperature compensation is superior to other algorithms in globality

which realizes high precision real-time temperature compensation. The experi

mental results show that the temperature coefficients of the scale factor

the full temperature zero bias range and the nonlinear are decreased from 141 × 10

-6

/℃

109. 111 mgnand 2223 × 10

-6

to13. 22 × 10

-6

/℃

9. 941 mgnand 294 × 10

-6

respectively after compensating with the temperature model

the experimental results prove the superiority and practicability of this method.