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。 In higher accuracy displacement measurements, the output of the displacement sensor needs to be temperature compensated. Adopting the multi-sensor data fusion method of BP (Back Propagation) network, the outputs of displacement sensor and temperature sensor are input into the fusion center as the input vector of the network and trained by BP network. Then the calibration samples are sent to the trained neural network to obtain More accurate displacement output. In order to overcome the shortcomings of the traditional BP network algorithm such as slow convergence and easy convergence to local minimum points, BP multi-layer feedforward neural network improved algorithm is used to compensate the sensor characteristics. The results obtained with MATLAB simulation are compared with the existing experimental data, In the same temperature fluctuations, displacement sensor output error is reduced by 3 times than the original, but also greatly saves time.