摘要
在较高精度位移测量中,需要对位移传感器的输出进行温度补偿。采用BP(Back Propagation)网络的多传感器数据融合方法,把位移传感器和温度传感器的输出作为网络的输入向量送入融合中心,通过BP网络训练,然后将标定样本送入训练好的神经网络,得到比较准确的位移输出。为克服传统BP网络算法收敛慢、容易收敛到局部最小点的缺陷,采用BP多层前馈神经网络改进算法对传感器特性进行补偿,用MATLAB仿真所得到的结果与原有的实验数据相比较,在相同的温度波动情况下,位移传感器的输出误差比原来的减小了3倍,而且大幅度地节省了时间。
Temperature compensation for output of displacement sensor is dispensable in high precision displacement mea- surement. Output of displacement sensor and temperature sensor is sent to multi-sensor data fusion center based on BP network, and then, calibrated sample is sent to the trained neural network to output the precise displacement. For overcoming the flaws of conventional BP algorithm such as slow convergence and falling to local minimum point, BP multi-level feed forward neural network algorithm is used to compensate the sensor property. Compared the experimental result with simulative result under the same temperature fluctuation, the output is much more time-saving and precise, which error is 3 times than it was.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2010年第2期90-92,共3页
Journal of Wuhan University of Technology
基金
湖北省基金(D2008279)
