基于PSO-BP模型的扩散硅压力传感器温度补偿

Temperature compensation of diffused-silicon pressure sensor based on PSO-BP

硅的电导率易受温度影响,导致扩散硅压力传感器的输出电压随温度变化产生漂移,需对该传感器进行温度补偿,针对该问题提出粒子群优化的BP神经网络算法(PSO-BP)。通过二维标定实验,利用温度传感器监测实验环境温度,得到扩散硅压力传感器在不同工作温度下的输入输出特性曲线,建立PSO-BP模型。该模型利用粒子群算法全局寻优的能力为BP神经网络算法初始权值和阈值选取最优解,弥补传统BP神经网络初始值随机选取的弊端,克服容易局部陷入极值的缺陷。实验结果表明,经过PSO-BP模型补偿后的输出零位温度系数和灵敏度温度系数均减小一个数量级,证实该模型能够有效降低温度对扩散硅压力传感器的影响。

The conductivity of silicon is susceptible to temperature,which causes the output voltage of diffused silicon pressure sensor to drift with temperature.Therefore,temperature compensation of the sensor must be carried out.To solve this problem,a BP neural network algorithm based on particle swarm optimization(PSOBP)is proposed.Through two-dimensional calibration experiment,the temperature sensor is used to monitor the experimental environment temperature,and the input and output characteristic curves of diffused silicon pressure sensor at different operating temperatures are obtained,and the PSO-BP model is established.The model uses the ability of particle swarm optimization to select the optimal solution for the initial weights and thresholds of BP neural network algorithm,which makes up for the shortcomings of random selection of the initial values of traditional BP neural network and overcomes the shortcomings of easily falling into local extremes.The experimental results show that the output zero temperature coefficient and sensitivity temperature coefficient are reduced by an order of magnitude after compensation by PSO-BP model,which proves that the model can effectively reduce the influence of temperature on diffused silicon pressure sensor.