摘要
为减少传感器非线性特性带来的测量系统误差,提出一种采用微粒群算法与BP(Back-propagation)神经网络相结合的方法设计误差补偿环节,将传感器非线性特性改造成为与实际物理过程相一致的不失真的线性特性,从而减小非线性误差。在电感微测仪位移测量系统实验中,采用PSO(Particleswarmoptimization)算法训练后,网络的BP算法收敛速度很快,且精度高,在经过120次学习后,误差平方E<0.001。
In order to reduce the measurement system error from the sensor nonlinear response, a sensor error compensation unit was designed by the particle swarm optimization algorithm and BP （Back-propagation） neural network. Using this method, nonlinear characteristics of sensor can be converted into a non-distortion linear model which is consistent with the actual physical process, and the nonlinear error of system can be reduced very much. In the inductance micro measured the meter moves in the measurement system experiment, after uses the PSO （Particle swarm optimization） algorithm training, network BP（Back-Propagation） algorithm convergence rate very quick, also the precision is high, after undergoes 120 times of studies, erroneous square E 〈 0.001.
出处
《电子元件与材料》
CAS
CSCD
北大核心
2005年第12期17-19,共3页
Electronic Components And Materials
基金
江苏省高校自然科学基金资助项目(02KJD510011)
关键词
电子技术
测量
误差
微粒群
神经网络
electronic technology
measurement
error
particle swarm optimization
neural network
