有限空间爆炸静态压力的温度补偿方法

为改善压阻式压力传感器的温度漂移特性，构建了基于遗传算法和小波神经网络的压力传感器温度补偿模型。针对小波神经网络收敛速度慢且易陷入局部最优解的问题，采用遗传算法对小波神经网络的连接权值、伸缩参数和平移参数进行优化。基于压力传感器的标定数据，分别采用BP神经网络、小波神经网络和遗传小波神经网络对其进行温度补偿研究，结果表明:遗传小波神经网络兼容了小波分析的时频局部特性和神经网络的自学习能力，表现出良好的收敛速度和补偿精度，经补偿后传感器的输出值更接近于标定值，其最大误差由?17.44 kPa变至0.38 kPa，最大相对误差由?14.0%变至0.38%。将该模型应用于有限空间爆炸静态压力的温度补偿中，取得了较好的实际应用效果。

A temperature compensation method for explosion static pressure in finite space

To improve the temperature drift characteristics of piezoresistive pressure sensors, a temperature compensation model for the pressure sensors was constructed based on genetic algorithm and wavelet neural networks. By considering the problems of slow convergence and high probability of the local optimal solutions of the wavelet neural networks, the genetic algorithm was applied to optimize the connection weights, expansion parameters and translation parameters of the wavelet neural networks. Based on the calibration data of the pressure sensors, the BP neural network, wavelet neural network and genetic wavelet neural network were used to study the temperature compensation, respectively. The results show that the genetic wavelet neural network was compatible with the time-frequency local characteristics of the wavelet analysis and the self-learning ability of the neural networks, showing high convergence speed and compensation accuracy. After the compensation, the output values of the sensors were closer to the calibration ones. The maximum error was changed from ?17.44 kPa to 0.38 kPa, and the maximum relative error was changed from ?14.0% to 0.38%. The constructed model is applied in the temperature compensation of explosion static pressure in finite space, and the practical effect is good.