一种基于局部多项式回归的气敏传感器模型优化算法

摘要

Accurate modeling of gas sensing sensor plays an important role in the accurate measurement of gas.Thus, a parameter calibration algorithm is proposed to solve the problem of unstable automatic system caused by gas sensor difference.Then, the relationship between the PSO-BP sensor response and the gas concentration is fitted by the combination of intelligent algorithm, and the two-dimension gas sensing character model is constructed.At the same time, an optimization method of regression analysis for the three dimensional modeling is proposed in order to reduce the temperature effect on mathematical model.Experimental and application results show that the parameters calibration method and the modeling methods improve the accuracy of the original model parameters and system reliability, and provide a high reference value for the preparation and application of gas sensor.%气敏传感器精确建模,为实现气体精确测量发挥重要作用.针对气敏传感器差异造成自动化系统不稳定问题,提出一种参数校准算法.然后基于改进PSO-BP神经网络拟合传感器响应值与气体浓度之间的关系特性,构建出二维气敏特性模型.同时为减少温度对数学模型影响,又提出回归分析三维模型优化方法.实验及应用结果表明,该参数校准方法以及模型优化方法,提高原模型参数精度以及系统可靠度,为气敏传感器制备与应用提供较高参考价值.