药品灌装线动态称重系统非线性校正的研究

由于传感器本身的线性问题和各种环境因素的综合影响,测量系统的输入输出总是存在一定非线性,因此对系统进行非线性补偿是提高系统测量精度的必要环节。以药品灌装线动态称重系统为例,由于传感器非线性造成的误差已将近2%。针对这个问题,提出了一种基于BP神经网络的传感器非线性校正方法,选用Levenberg-Marquardt反向传播算法的训练函数和感知器权值和阈值学习函数可以将系统的非线性误差控制在0.4%的范围内,从而大幅度提高药品灌装线称重系统中传感器的测量精度和工作效率。该非线性校正方法也适用于其他的测量系统。

A useful nonlinear rectification way based on BP neural network is proposed in this paper.Using the trainlm and the learngdm,the nonlinear of the weighing system used for the drug fil ing system is control ed in the range of 0.4%. Thus the efficiency of the entire weighing system can be improved effectively.Furthermore,this nonlinear rectification method is also applicable to other measurement systems.